I would really hate to see anything bad happen to GCP after the changes in leadership.
Based on my own experience as the CTO: After being an Azure shop for a year we've migrated ~50-100 VMs to GCP and I love the GCP products.
GCP is:
- Simpler to use
- More tailored to people with Linux environment
- Leader in K8S
- Has good support
- So much cheaper (in our case we saved ~60%)
- Has great UI and understandable primitives.
My only pet peeve is the fact that exporing your spend is practically impossible unless you're a BQ guy that can work directly with report exports.
PS: We're building our future infra on K8S to allow us to migrate more easily to a different could if something goes awry with GCP, I really hope there won't be a need to migrate back to Azure and its arcane and high pricing, strange UI, worse tooling...
For anyone else who is a Google Cloud Customer. I 100% recommend "Exporting Billing Data to BigQuery" [1]. This takes less than 5 minutes and will give you near real-time extremely detailed billing data into BQ where you can explore it via SQL. Second, generate a real-time report off this data with something like "Visualize GCP Billing using BigQuery and Data Studio" [2].
That blog post is good, but it is over built in my opinion, all you really need is a single daily sum of all spend across your account and chart that. Maybe a stacked column with product color or something. This will allow you to quickly see if one product consumption is shooting up unexpectedly at a glance vs checking something at the end of the month and working with your account reps to figure something out.
6k z z
5k z z
spend 4k z z z
3k z z z
2k x x x x x x x x x
1k x x x x x x x x x
date 1 2 3 4 5 6 7 8 d+
x = expected
z = unexpected
For example, maybe you left a test env running or something and all of the sudden you're spending an extra few K per day. Maybe you're doing some query loops over a large BQ dataset and racking up tons of $$. Put the chart up on your big TV screen in the Ops area so everyone can quickly check it with all your other metrics. When you're spending 100k+/month, with a few folks digging around the console, it's hard to notice an extra 5k-10k here or there. This quick tip will totally help you visually sanity check if something is off. So, this might not be a killer solution for folks spending a few thousand a month but will totally saves tens of thousands as you scale up.
Couldn't agree more. GCP has focused on building the core services right and making everything work together seamlessly. I would hate to seem them go in the direction of AWS/Azure where they build a zillion different "products" but every one is only 80% of great.
I disagree that GCP offers better core services than AWS and in areas like Data Science or Serverless GCP has almost nothing to offer.
And one of the advantages of having a zillion different products approach is that you can trust AWS to grow with you and improve their products over time. Same can't be said of GCP.
Compute is compute. Both providers offer similar instance classes for the most part. GCP offers sustained-use discounts, in addition to committed use discounts that match up to reserve instances, plus billing by the minute.
Spot instances can be hugely variable, but cheaper than prremptible. Preemptible offers predictable costs with a guarantee that the instance will go away every 24 hours (and come back).
Network IO is way better.
Inter-region networking comes for free.
GCP ssd persistent disk lands between AWS GP2 and IO1 and gives way more predictable performance.
Local disks can be attached to any instance and end up giving similar performance and being way cheaper than I2 and I3 instances.
Compute is just compute, but if all you care about is running VMs, and not taking advantage of the managed services, what is using Azure/AWS/GCP buying you? You’re not saving money on server management, you’re not moving faster and you’re spending more than on a colo.
> with a guarantee that the instance will go away every 24 hours (and come back)
There is no guarantee that the instance will come back. And it can be shut down in under 24 hours. Not that that's a problem compared to AWS or Azure.
>Because preemptible VMs have no availability guarantees, you should design your system under the assumption that any or all of your Compute Engine instances might be preempted and become unavailable. There are no guarantees as to when new instances become available.
> I really hope there won't be a need to migrate back to Azure and its arcane and high pricing, strange UI, worse tooling..
Azure's VMs are definitely overpriced compared to both GCP and AWS, but most other services are reasonable.
Not sure what you mean about Azure's "strange UI"? Personally, I find the Azure Portal to be far more consistent than AWS' or GCP'. I also think it looks and feels great, and I'm obviously not the only one - I literally have customers asking me to replicate the Azure UI for their web apps!
Lastly, I find it somewhat incredible that you would complain about Azure's tooling - I've found it to be excellent: a snappy web UI, cross-platform Powershell, cross-platform Azure CLI, REST API, ARM templates... seriously, I'd be interested to hear what your beef is with the tooling?
I like the look of Azure as well, but getting something done is such a chore. Eg: firewall.
You need to drill down to the network card primitive, figure out which subsection you need to configure and applying the same rule is again an exercise in clicking.
In GCP:
Apply a tag.
Create firewall rule that targets the tag.
But then again, this is higly subjective and I do understand that somebody would prefer Azure way.
Re: tooling, when I was using CLI was in both Node and Python, docs were lacking and the strange Resource Manager vs. Legacy had so many quirks we had to run windows to get proper powershell for configs of VMs.
Leader in k8s? Not for production. I defy you to do a gcp gke deployment without using beta or alpha features (that specifically say don’t use this for production).
Stackdriver error reporting springs to mind. And working with SSL certs/let’s encrypt is a close second.
If you want a couple more peeves for your list how about documentation? I regularly find documentation heavy on theory light on specifics.
GCP is my favorite thing that is almost good enough to use.
I’m hoping the change of leadership actually improves things. GCP could stand to re-focus on a “customer first” philosophy.
I feel like their support has really been improving. 4 years ago it felt like it was non-existent. Been pretty happy with it the last few times I have had to contact them.
I don't think we use any Alpha features, I know there are a couple Beta APIs from Kube we use, but we would be using those no matter where we ran our Kube deployment. We use fluentd for our log collection and Prometheus for metrics and alerting combined with Elastalert. So don't have any experience with Stackdriver.
Why is working with SSL certs an issue? We use the NGINX Ingress with cert-manager and have yet to have an issue. We only have a couple domains pointed at the cluster though, not sure if thats where your issue comes from. The two just worked though with very little tweaking.
My team is running a couple thousand cores on GKE and have been in production there for a year. These are multi-tenant clusters with several dozen deployments of my company's application. "production" tenants get automatically configured SLA monitoring via Pingdom which it's a health endpoint every 30 seconds and has something like a 30 second timeout.
Average availability for the year across those environments is 99.99%.
I'd say that's decent enough to use.
Their support is hilariously bad at times, though.
Lack of IPv6 support is what keeps us off GCP or AWS, HTTP fronting is great, but when it comes to realtime interactive streams to mobile devices, the ability to save significant latency and avoid the hot mess of CGNAT is invaluble.
IPv6 is not a first-class protocol on AWS, but it is absolutely doable for most scenarios. It still requires having something of a dual-stack network in place (you definitely can't have a v6 only network of anything meaningful).
I find it a bit ironic that the mobile and web side of the house are pushing very heavily for IPv6 adoption and yet they expect their cloud customers not to need it.
We're on OVH for most things that aren't latency sensitive. For things where every millisecond counts for quality, we either find a 2nd tier cloud provider with decent peering to most of our end users in that area, or less common colo or rent a server (some areas just don't have 2nd tier cloud providers).
For me support is:
1. Reasonably priced (especially now, with role based)
2. Fast
3. Helpful
That was my experience when we needed something.
Dont get me started on docs and operations.
Eg: Using Azure we found out the hard way that the Firewall was cutting connections after 5m of inactivity. Changing that setting? Their docs are wrong, stack overflow kindof helps, but to fix it we had to use windows based tooling in a vm since their linux based stuff didnt work.
I know this will happen sooner or later with anybody, but on average the experience with GCP was superb.
I do agree re: certs. That is the only thing I really hate, doing LE when they could run this for you. But they will launch an SSL solutio since they have one now for AppEngine.
>"Eg: Using Azure we found out the hard way that the Firewall was cutting connections after 5m of inactivity."
Could you elaborate on this? Which firewall is this? I remember something similar with their load balancer but I thought this was fixed years ago. Might you have some link to the issue? Thanks.
Amazon has nothing in the category of exception tracking.
StackDriver error reporting is a lite product intended mostly for folks with small applications on AppEngine.
There is a recent post by Corey Quinn entitled CloudWatch Is of the Devil, but I Must Use It discussing the extremely poor usability of Amazon CloudWatch [0]. Here’s another story about the general weirdness of the product [1]
I would strongly recommend using a dedicated exception tracker like Sentry.io, they have an excellent full featured product at a extremely reasonable price.
GCP has clearly documented Platform Launch stages for Alpha, Beta, and General Availability [2].
The Kubernetes project itself follows the same definition for these identifiers as GCP.
Specifically Beta is considered to be acceptable for most production use-cases, but it is not guaranteed that the API will remain backwards compatible.
So if you’re running the majority of cloud services, beta is fine. If you’re running a system that will cause permanent harm to humans, don’t use beta.
One last note, GCP has just released a fully managed certificate product for their global load balancer. There is fully integrated Kubernetes support which has not yet been announced formally [3].
Certificates for Kubernetes on Amazon are less integrated than the above. There’s several options, it’s not clear without extensive research and testing what the tradeoffs are, and in the end it turns out that there is not an officially supported Amazon option.
Azure’s main feature is that it’s not connected to Amazon retail and they’re committed to Windows. So their clients are often in it for the long haul good or bad, those who can switch do.
What’s the benefit of being “not connected to Amazon retail?” Serving a large customer that demands scale seems useful for driving the platform forward—what is the downside?
If I had to guess it's the perception that Amazon is out to compete in any and everything, and if they wanted to they could figure out what products and services are the growing the fastest based on the customer usage data from AWS and then come out with their own competitor.
I used to do $100,000 a month in revenue on Amazon with one HP skew. The product grew to #1 in its category within a couple months, another month after which Amazon started selling it to, "shipped from sold by Amazon".
Amazon is happy to help be your growth partner until you prove worth devouring.
Why would Netflix fund Amazon's prime video service?
Why would Walmart fund amazon.com and Whole Foods?
Why would Dropbox fund Amazon Drive?
AWS is their most profitable business in terms of margin. It makes sense to avoid funding your competition if possible, which is a problem for Amazon given how many business divisions they have and continue to add.
Competition. As Amazon grows as a conglomerate, competitors are less inclined to use their services, even if that one service isn't what they're competing over.
GCP is reflective of Google's internal architecture, just as Azure and AWS are reflective of Microsoft and Amazon. This isn't surprising, they're leveraging the same foundation they used to launch their flagship products in terms of infrastructure, knowledge, and culture.
Google is very different internally from the enterprise loving Microsoft, and they're more technically fluent than the customer focused Amazon. Googles edge targeting and process scheduling are also very different.
These differences make GCP a very different product than it's competition. Azure is essentially leveraging the ecosystem fracture that Microsoft already holds, and the fact that Microsoft knows how to speak the slow drawl of enterprise. GCP is situated to solve things the Google way, which isn't the way most people run their infra.
Yes, yes. The Google “way”. Because everyone has the same issues google has and wants to solve them in the same manner. /s
Google lost the Cloud Train. They need to focus on things that work and I would not be surprised if GCP was not a thing 5 years from now.
A move that was necessary. Diane wasn't able to shift the existing Google culture to one where the enterprise customer's needs came first. (Observable as reflected in their pricing, sales, and support challenges.) Thomas now has a similarly tough job navigating the organ rejection risk of transplanting anything that looks like Oracle culture and practices.
> Diane wasn't able to shift the existing Google culture to one where the enterprise customer's needs came first.
Citation needed -- from my knowledge, the culture shifted dramatically -- she did succeed, and now that's done.
Granted, I'm referring to Google culture within Cloud/GSuite... obviously she didn't affect culture at YouTube or Search because those weren't her areas.
This is completely incorrect. Under Urs'es leadership it was hard to work on enterprise features due to explicit direction from him "we don't support enterprise because it's too hard". Diane completely changed that, developed coherent GCP/G Suite strategy, and now GCP organization is very enterprise-oriented. Hiring of Oracle dude kinda re-iterates this.
Sure. But they are in 3rd place. And so very, very far behind AWS in the range of products, quality of service, developer ecosystem, number of regions etc. And while GCP is still focusing on the basics AWS is locking developers into their platform with paradigms like serverless.
And once IBM launches their cloud I would actually put them ahead of GCP given how useful their Compose suite is.
IBM already has a cloud called Bluemix, and they bought/merged Softlayer into it. It's not great.
Compose is just a managed database service which IBM also acquired. I don't see how that alone would put them ahead of any cloud, especially when they're already lacking in infrastructure and software.
They also just acquired a tiny little company called Redhat who I am sure knows a thing or two about infrastructure and software. And Compose is the most useful database offering on any cloud IMHO.
Redhat is tiny compared to the clouds. Softlayer is aging and IBM does not have billions in infrastructure buildout to compete. Redhat can provide software but definitely not enough to put them ahead of GCP as you claim.
And Compose manages databases on AWS and GCP too, similar to Aiven and others, so I don't see how that alone makes IBM special.
IBM's capital spending for cloud infrastructure is a drop in the bucket compared to the big 3 cloud providers. Same with Oracle. They aren't going to be able to compete for big dollars.
But I will have to admit that AWS seems a lot more committed to their cloud products than Google. New GCP products and features take forever to be introduced or phased out of beta.
For example, cloud functions entered beta some 2 years ago and went out of beta a year later. Even today it's only possible to use Node and Python. Only recently the Node runtime was upgraded from old Node 6 to Node 8.
During that same time, a small company like Zeit has created a number of complete cloud products for developers. Not only that, but Zeit Now v2 is better in many aspects than Google Cloud Functions.
Why isn't google invested in GraphQL? They could have come up with something like Prisma with all their talent and resources.
I agree, I think for most people that's the main hesitation with Google Cloud. Especially when there are outages, it feels as if Google has very few people working on the product.
It is always worth pointing out, since few people are aware of this, that the servers you rent from Google Cloud do not accept any network packets from US-sanctioned countries; if you are ideologically inclined to the idea of an internet with few borders, free software and so forth, then Google Cloud is arbitrarily crippled infrastructure.
In any case, people should be aware of those limitations when choosing a provider.
The Cuban citizens that have access to the internet do so through Nauta, a government ran network of wifi hotspots. This network is heavily filtered and operates basically on a whitelist.
Even if GCP accepted packets from Cuba, your problems probably would have persisted on the Cuban end too.
Neither AWS nor Azure have this kind of block. Nor does Digital Ocean or others. Neither are Google Products such as Gmail blocked.
It's ironic that it's fine for Google to provide it's products in those countries, but not for me, if I host my app on their cloud.
US sanctions do not require an infrastructure provider to do this, much like they do not require a broadband provider to filter packages destined to those countries.
Can anyone help me think this through: it seems Google/Alphabet shouldn't just be #3 cloud provider, but the cloud itself. It should have been difficult for consumers to separate the concepts of cloud/Alphabet, as cloud is so integral to the Alphabet business.
Instead the gold podium is occupied by a book re-seller. But the entire Alphabet product line, from the amazing distributed Big Tables to the stitching of 3rd party satellite polygons, is just a front-end cloud use case.
So why didn't it happen? Departing CEO Greene writes in the OP link that she was only supposed to be running GCP for 2 years. Why on earth was an interim CEO there in the first place
Google missed cloud in the same way Microsoft missed internet. Google entire product strategy is defense of Ads/Search revenues and napalming any emerging threats. Through this perspective when cloud technologies started to emerge, they didn't seem important. Then it became clear that cloud providers grow and consolidate so fast, that Google is not going to be hardware buyer #1 anymore and will lose access to best hardware discounts, threatening profitability of Ads/Search itself!
This was pivotal moment: backwards Cloud PA nobody cared about (fun story: Cloud PA was once forgotten at company-wide key result meeting, it was fun to see how couple hundreds of people watching video-cast in the auditorium looked at each other and like "meeting is over, they didn't mention or PA once, what is going on!") suddenly became the priority, was merged with TI (Tech Infrastructure) and everything became "Cloud! Cloud! Cloud!".
Google's search foundations were built on a private cloud, in contrast to the prevailing architecture of that time. Google never denied the utility of building abstractions over commoditized machines the way Microsoft denied the heterogeneity and decentralized aspects of the Internet.
IMO, Google was just too focused on solving problems for human beings vs. outside institutions; that's why they neglected to invest in making Borg into a public platform.
Borg can’t be a public product. Look at the way people complain how hard it is to use EC2 then multiply by a large factor. No isolation, sudden death, zero local I/O resources. It’s technically marvelous and dirt cheap but the masses aren’t ready for it.
1. Google saw their infrastructure as being a key competitive advantage, not something to be commoditized.
2. More importantly: Google's core business has ridiculously high profit margins, which made them institutionally incapable of seriously pursuing low-margin business opportunities.
If google starts "dogfooding" and moves all of their compute/storage, etc to being purchased from their cloud division, then they would be the worlds largest cloud vendor the next quarter.. That would be pretty interesting to see..
Google acquired her company without any real product for $380MM. She cashed out, now she is gone. Simple as that. It's going to be harsh ride from here with no upside.
Diane was already a billionaire prior to the acquisition. She also donated all of the Bebop money to charity anyway[1]. And she’s retaining her Google board seat.
This is my guess: Doing enterprise requires significant amount of 1:1 contact. You need thousands of people dedicated to each of the 50,000 or so top accounts. These folks do site visits, gather info on legacy stuff, try to figure out your requirements, connect to dev/PMs from core team, help you do sizing, convenience CTO et al about benefits, take your phone call at 2AM so on and on. Doing enterprise is mainly about creating, training and managing vast sales force in financially efficient way. I suspect GCP hasn't focused on this part much. It's very capital intensive operation and if not done right it can drain your balance sheet very quickly. For enterprise related businesses this is about the singular skill for an organisation that matters.
You seem to base #3 on false numbers. AWS is the first huge player and they have a huge advantage due to it. Companies like Microsoft/Oracle/IBM have huge enterprise presence so it does skew numbers on the cloud. Competency should be measured on fine grain. As a GCP customer for 2 years and having evaluated all major providers i can say this without doubt that #3 is not the real standing by no measure.
Book re-seller. Hahaha. May have been true at some point but right now it’s THE CLOUD PROVIDER.
The only other player in the cloud is Azure. Everyone else is struggling to stay relevant.
I did interview with Microsoft Global Network Services several years ago (before Satya Nadella) and got a sense on the infrastructure and money they were pumping into Bing.
They were replicating the style of data center deployments that Google was known for in terms of scale and efficiency. I wasn't really privy to any specific details beyond the Clos architecture they'd adopted for their underlay and the fact that they rolled their own switches and SDN.
I don't recall specifically when this was, but I do recall being impressed that they had something deployed and working. The only folks anyone were certain had done this at the time was Google. Facebook would later announce something similar with their Wedge switch and now there's a fully baked OCP standard.
They point of all this I suppose is that they were heavily investing in scale for Bing and Office 365. I suspect the product of all that work is what became Azure.
TK has a reputation for bad temper. I have heard senior execs at Oracle say that they take great care to ensure that star engineers don't come into contact with him directly, lest they lose those engineers. Would be interesting to see how this plays out at Google.
There's a great episode of "Masters of Scale" with Diane Greene where she talks about her time at VMWare and then about how she ended up taking the job of leading Google Cloud. https://mastersofscale.com/diane-greene-look-sideways/
Oracle and TK understand one thing that GCP does not: the importance of great support. Oracle charges a ton for top-tier support, but if you have a problem they can't solve remotely, they'll send a team of engineers to your site and work night and day until it's fixed. In extreme cases, they'll duplicate your hardware and software setup at their expense and run in parallel until they can duplicate and solve the problem. That is serious enterprise commitment.
Google has no cloud counterpart to that level of service. If TK brings an equivalent orientation that makes customers believe that GCP will do whatever is necessary to resolve a problem, he'll have contributed significantly to its success, IMHO.
> In extreme cases, they'll duplicate your hardware and software setup at their expense and run in parallel until they can duplicate and solve the problem.
I knew there must have been a good reason they overcharged everyone for everything...
In all seriousness, I imagine this is the type of thing you can do when you have (or traditionally had) close to a captive market for specific things (government contracts) that allows for massive profits. It's easy to throw a few tens of thousands of dollars at black swan support events so you can point to those as examples of what you're paying for when people point out that you've been repeatedly found to violate contracts and fraudulently misrepresent yourself.[1]
Google trying to grow a cloud counterpart for that level of customer service will be interesting to watch. If one tried to parachute a gaggle of Googlers into a company's site to make their systems work with Google Cloud, I imagine the end result would be the company's existing infrastructure being entirely rewritten from scratch before any changes were made to the Cloud infrastructure.
Oracle Cloud Infrastructure is a totally different approach, for what it's worth. Feels no different from working at AWS etc. (other than OCI gives a damn about operational burden and paying down technical debt, rather than letting the fires keep on burning as long as the features are being pushed out)
TK is very much old school Oracle though. Doesn't feel like a good cultural fit with Google.
I agree completely, having worked at a few Seattle based cloud providers. Perhaps TK’s personality did not fit with the new direction of Oracle, centered around OCI and its ethos.
Notwithstanding that they are 3rd in terms of revenue, GCP has gotten to a great place in the last few years. The products are fantastic. Great docs, super consistent APIs and products, great pricing. Really a joy to use. The bet on GKE really paid off too.
During her tenure it moved from something I wouldn't even consider using, to first choice for my new startup.
While I agree their offering is quite good, I understand why enterprises do not want to invest on it. Some reasons that come to mind:
- Much less people with proper platform knowledge (not even talking about certified).
- Not much trust on deprecation policies (AWS keeps running very old and deprecated services virtually forever, there’s no guarantee GCloud would do the same given Google history).
- Not from personal experience but customer support is not top notch as with other providers from what I have read and heard.
In my opinion those three reasons alone are big flags for many corporations, which might prevent them from getting the big contracts.
Re: deprecation, I deployed a personal project 7 years ago to AppEngine, and it has run continuously since then, costing me a few cents each month above what the free tier provides. Since then I've redeployed a few times with little bugfixes or features, and all the infrastructure changes together have required a few lines to be removed from config files, and that's it. It still runs exactly as designed, 7 years later. By far one of the best effort/reward ratios I've had on any platform. No OS upgrades, no package version conflicts, no downtime (that I've noticed), just deploy and forget.
That said, I'm not sure I'd use it for anything that might scale heavily with datastore writes. They get expensive very quickly if you can't cache and group the writes.
On top of this, there is the not-so-distant issue with the sudden price hike of Google Maps API, which made many companies struggle (in this case especially small ones).
The maste/slave datastore deprecation burnt us. The migration couldn't be staged and when migrating the db locked up. The issue had to be fixed on googles end and caused us a bit of downtime.
The engineering resources to migrate was also a problem with the whole stack having needed to be updated.
What is "proper platform knowledge"? Certifications are generally far inferior to hiring actual good engineers, which in my experience Google does. Comparing AWS to Google, our Google support is way higher touch and they give us way more experienced people (actual engineers).
Most enterprises don't have a clue about computing though, so I can buy that they would prefer bling like certificates to getting a contact who is a real engineer.
I think if some Microsoft, Amazon, and Google employees didn’t like my business for whatever reason and pressed their leadership to kick my company off their platform, Google would be far and away most likely to give in and close the account.
I think its really #4 - cloud is still a minor side bet for them, something investors are expecting them to try to work at but that the core company doesn't really care about. its kind of a bigger google X project. aws and azure are significant company bets from amazon/microsoft where they are staking their future, google cloud is not a bet on googles future.
A VMWare founder couldn't force Google culture to be enterprise-y enough, now an Oracle exec will try that ... If he fails, i guess the next guy will be a 4-star Marine general.
As far as i heard, the VMWare people Diane brought into the Google Cloud org were screaming from the pain of being an adapter between the hard place of the rest of the Google (i.e. the infrastructure, etc.) and the sledgehammer of enterprise customers who, still mentally being unenlightened dwellers of the Dark Ages, fail to understand the 30 seconds shutdown notice suddenly coming from deep inside the Google infra guts as an "advanced maintenance notice".
Lets see what happens to the Oracle guys who are already coming upon the Google Cloud like Viking drakkars upon Northumbria shores.
This is Google's problem. Total and complete lack of long-term view. Just look at how many projects they shut down every year.
Do they really think they were going to become market leaders in three years? Come on. The market for cloud infrastructure is in its infancy. The opportunities are there. But you don't build a large business within a potentially multi-trillion dollar industry by being impatient.
And hiring a former Oracle executive just shows how impatient they are... It a complete change of the playbook and it likely means that they are scrapping a huge chunk of Greene's original go to market strategy for Google Cloud.
There is a difference between their consumer products (which have a terrible reputation) and Google Cloud which does not have the same history and is well-supported.
However there is that Maps pricing change that should be noted, and perhaps counts the same as forced deprecation for many clients.
The Maps example is real. However in their confused current branding Google Maps Platform is outside of Google Cloud Platform yet within Google Cloud, alongside Chrome, Android, and G Suite.
They haven't done any pricing change like that, nor anything else effectively equivalent to forced short-notice deprecation, for generally available Google Cloud Platform services. (The famous Google App Engine pricing change happened upon GAE's departure from preview.)
Hopefully the new CEO can disentangle the horribly messy branding conflation of orgs which work rather differently.
Google is now effectively a conglomerate, even ignoring Alphabet, maybe even multiple conglomerates within Google. The impacts of their org chart on their users shouldn't be this opaque to people who haven't worked at Google. (Disclosure: I have worked there, though not since 2015.)
There may be a difference but they’ve tied their cloud branding so close to the Google brand that the reputation of Google consumer bleeds over completely, good and bad. I’d say they made a mistake in not having a clear branding differentiation from the parent company like Azure and AWS do.
Apigee API BaaS,
Fabric,
Google Search Appliance...
However, I was referring more to Google's general tendency to shut down projects across the board. The list of consumer products shut down by Google is pretty long (Google Reader, iGoogle, Aardvark, Google Inbox, Orkut, etc..)
None of those are part of Google Cloud. But really; I don't see a CTO making an informed decision on which cloud vendor to use by the fact that they liked Google Reader and Orkut 10 years ago. Just like I would not expect a good CTO to base their decision to not use AWS on Amazon discontinuing the Fire Phone in 2015.
Fair point. You're right about consumer product shutdowns not being a good reason to disregard Google's cloud offerings.
I think both Fabric and Apigee API BaaS where indeed Google Cloud offerings. Google Search Appliance if I'm not wrong was part of something called Google Enterprise which eventually folded into GCP.
GSA was a physical box you stick in your on-prem data center. It's about as un-Cloud as it gets, and IIRC its deprecation predates Google Cloud as a brand.
>>>> Which google cloud projects have shut down this past year? Or Ever, for that matter?
>> Apigee API BaaS, Fabric, Google Search Appliance...
Curious, what usually happens to the engineers on projects that get shut down? Do they get redeployed to other areas? What if the team is in a small office?
I've walked past the Apigee office in Reston, VA...what would happen to those engineers...i imagine there arent many other options for redeployment short of moving cities...
Those engineers shop themselves around and maybe move offices. Google often destaffs launched, current products so this is not too weird of a situation for a Googler to be in. It’s not something that happens only to canceled products.
Isnt it a little hypocritical for you to say this? I went through your profile and saw that you were involved with Contactive (https://news.ycombinator.com/item?id=8096597) which doesnt exist anymore. I'm assuming that doesnt exist today because of very good reasons. Same for other companies as well.
Yes, I'm a hypocrite for having worked for a company that failed to find product-market fit and having the nerve to give an opinion about why I believe a company like Google should be more consistent with their product offering and the long-term vision of their ventures.
Thanks for pointing out my hypocrisy and why I don't have the right to have an opinion about anything anymore.
I also checked your profile and you're pretty good at defending everything Google does, but I don't judge you for having those opinions, nor I think you're a hypocrite for thinking in a certain way.
For the record, I upvoted you and I think its important everybody should read what you wrote. As you mentioned, companies need to be consistent with their product offering just like how we expect people we interact with online to be consistent as well.
If I criticize Company A for doing something, and goes around and does the same thing then it'll be very hypocritical of me.
The product-market fit you mentioned doesnt just happen for startup, but for big companies as well. A dollar is a dollar everywhere you go. Just like you found it was OK for your company to sunset a certain product, big companies also decide to do that.
Also, please dont put words into my mouth. I didnt say you dont have a right to say things. You do. But when you do, and please be open to replies.
Yes. Sorry for the angry response. I'm totally open to replies and debate. I just wanted to emphasize that I was merely giving an opinion and that I didn't think it was fair to characterize my comments as hypocritical.
Companies have the right to sunset products if they're not profitable. Of course, they do. I think the problem is that in general people have high expectations from companies like Google, Amazon, and Microsoft. I certainly do, and that's why I think Google should be more committed to a long-term vision in their Cloud business.
>This is Google's problem. Total and complete lack of long-term view.
Diane Greene only planned to be there for two years, so three years in, this change is overdue. If anything, keeping her longer demonstrates the opposite of what you're saying.
Google is an advertisement company and their long-term view is getting revenue from showing you ads. The rest is just the means of maintaining the "we are not evil" image so people won't get scared away to alternative search engines. In that perspective, making constant buzz about new projects and closing them once the hype is over makes full sense.
Google Cloud Platform is a separate company/division focused on cloud computing, which has potential to be a trillion-dollar industry. Alphabet/Google could possibly make more money from this than ads in the future.
I wish Diane well in her future endeavors and I certainly agree it would be great to see more women in the C suite.
I would be curious to know if her passions are intertwined with her departure, specifically if she felt that Google had structural issues with women in leadership roles. She is essentially being fired as the lead of this organization and being replaced with a fairly prototypical male lead. I am not implying that Thomas isn't qualified, his LinkedIn profile suggests he have very relevant experience, Diane also has the experience in her resume as well. What will help Thomas succeed in this role?
The bottom line is I would like to see more women in leadership positions and I am sad to see us lose one. I would like to understand (although I realize that isn't possible) if her departure was preventable, and if so what would have had to be different.
>Diane also has the experience in her resume as well.
Nothing in her resume actually suggests she had the experience for what Google needed. VMware became popular in a similar way to Redhat, in that it solved a technical issue nobody else was solving at the time. It didn't have explosive growth until EMC took over and with it EMC management style/structure. She was just along for the ride the second the company was sold to EMC.
>What will help Thomas succeed in this role?
He was in a position of power in what is one of the most successful sales organizations in the history of tech. I don't like Oracle, in fact my post history will show that I loathe them. But their salesforce has absolutely dominated the tech industry for 2 decades+. Google needs someone that knows how to sell to enterprise, they don't just need another nerd at the top. Based on my experiences with Diane she was an extremely gifted technical talent and an extremely poor saleswoman and leader.
> until EMC took over and with it EMC management style/structure. She was just along for the ride the second the company was sold to EMC.
This is not true. Diane was the founder of VMware. VMware is a virtualization technology that runs independently of underlying hardware and infrastructure. It could succeed only because it was able to convince its customers and partners that it is neutral to all forms of infrastructure, whether it is compute, networking or storage related. Most of said partners - companies as IBM, HP, Netapp etc. - are direct competitors of EMC.
This is the reason that EMC allowed VMware to operate independently, be listed on the stock exchange as a independent entity and have its own operating culture. And Diane was the main reason that VMware maintained this kind of independence. Source: I was a VMware employee.
Diane was a founder of VMWare. VMWare was very successful at getting Enterprise customers to convert their massive single use servers into "Liquid IT" where VMWare instances ran the world. NetApp, when I was there, was selling a bunch of NAS into clouds of servers running VMWare. Then she left VMWare and started BeBop, a company that was dedicated to getting Enterprises to move their applications into virtualized compute clusters. When she was acquired/hired by Google is seemed like a slam dunk that she was the person who was going to make Google Cloud the premier service for Enterprise companies to host their infrastructure on.
As I read it, that was very much in line with where Google saw Google Cloud going.
Then we see from the market analysts that Microsoft has (according to the analysts here, I can't really vouch for them) edged Google into third place in the 'Cloud' market by capturing Enterprises moving their applications to the enterprise space.
Then we see Google replacing their head of cloud with a high power sales guy.
And as always, "reading the tea leaves" as an analyst might say, is always an inexact science. So folks outside the situation (like myself) can't know what is really going on.
> Based on my experiences with Diane she was an extremely gifted technical talent and an extremely poor saleswoman and leader.
This seems to be a pretty damning statement, as someone who has direct experience with Diane, what did you see that she could have done differently?
GCP doesn't need to figure out how to sell to enterprises.
They need to offer a better product with more features, support etc. Because adoption within the enterprise isn't driven by the CEO or CTO. It's driven by individual architects and developers.
YMMV widely. It's not at all uncommon to have large software and infrastructure deals signed by CTOs and Business Analyst type people based on advertised product features and compliance certifications, without ever speaking to architects, engineers and implementors about technology fit, compatibility, etc.
Most non software firms consider technology as a cost center. Most don’t care about the technical merits of the platform; all they care is cost and support.
This is why IBM cloud is seeing adoption despite not being as full featured as aws or gcp.
Okay, you are correct, let's not say she was fired, she was separated from the company.
Separations can happen for several reasons and they have different terms, I recognize that most people associate "fired" with incompetence or misconduct and that likely wasn't the case here[1].
Generally, with senior leaders, they have a set of objectives that they are tasked with working toward, and every year is a negotiation over whether or not they are meeting their objectives. When they aren't meeting those objectives (or perhaps enough of those objectives) then there is a discussion about next steps. Which is a euphemism for "this isn't working for us, we need to replace you with someone who can meet these objectives."
Now is that getting fired? or laid off? I think it's semantics. At the end of the day you don't have a job with the company any more. Few senior leaders that I've interacted with have ever called out the company for telling them they were going to lose their job. If you have reached that level you should have the maturity to understand that it is a fact of life that this happens. Scott McNealy used to send congratulations to people who were promoted into the senior ranks with "One step up, one step closer to the door." That reflected the reality that there are few alternative positions within a company for someone who is leading a big chunk of the company.
When the separation is the idea of the employee (which is to say they quit), the notification pattern seems proactive on the employee's side. A press release that the employee is moving on and that the company is working on finding a successor, and then sometime later we get the "I'm actually leaving now, and here is my successor." press release. But there is no hard and fast rule.
Bottom line is that the term 'fired' has the connotation of malfeasance and that doesn't seem to be the case here, I should have used "likely involuntary separation" but that seems a bit wordy, although it avoids the baggage of the word fired.
[1] As we can see from documents about Andy Rubin's departure, all of the press at the time read like Andy was just moving on to bigger and better things, when in fact he was being separated from the company because of credible accusations of sexual harassment. My point being that we on the outside can't know why a senior leader is leaving really, the copy in the press releases is carefully crafted regardless of the actual reasons.
> The bottom line is I would like to see more women in leadership positions and I am sad to see us lose one.
Absolutely. But let's be real, the odds are likely that Google tried to bend over backwards to make it work. They want more women in those roles and generally.
I hate to be a negative nancy, but this is a huge blow to GCP. People really underestimate how badass she is/was. I really worry that GCP will really hurt Google Morale since they were so sure that they would win this market.
"Those the gods wish to destroy, they first make arrogant"
Google’s Cloud weakspot is Sales and Support - the handholding to get clients over the finish line. The new hire should be able to layer this on to Google’s existing technical strength.
In my experience Google has awesome sales people, but the ones I've spoken to have an engineering background. Convincing other engineers that GCP is technically superior isn't that hard, but (unfortunately) engineers aren't usually the decision makers.
People expect Google to work, as Google trained people to expect its things to just work. Like Google.com, Gmail, Android.
And GCP is complex, it will break down. And then people just get frustrated. Yes, it works from the CLI, but fuck that. I know AWS is slow and expensive, but at least it does the basics. GCP was unable to launch VMs - for fuck's sake - on the UI just a few days ago with mysterious resource not available errors. (Anecdotal first hand experience, I've spent at least 1-1 hour trying to get an instance in a EU zone, to no avail. Even waited half a day between.)
GCP is great but we couldn’t find a way to get logs to go anywhere but to StackDriver and it being so expensive we had to jump ship because of it. Basically try as we might all the pods that ran our containers had stderror and stdout hijacked by StackDriver
You mention pods so are you talking about GKE clusters? If so, Stackdriver integration is just another service installed when you setup your cluster (using fluentd as the agent). There's also an updated beta version of this monitoring, but you can disable all of it easily in the GKE console with a few clicks.
Container output isn't hijacked, it's all logged to files, and those files are tailed by the logging agents. You can run multiple logging agents, for example we use https://logdna.com/ (highly recommended) and it works fine in addition to stackdriver logs.
Ok this might be embarrassing to admit but when we used GKE about 8 months ago we tried to do this but we couldn’t get our own logging to get anything from the pods. Maybe the tech lead was biased and wanted to go to AWS all along. I’ll have to take another look at it because I liked the service. What I didn’t like is that I on many occasions saught help from Google but got no response. We didn’t pay for support but just an email a few days later from support showing the methods to disable it and maybe a look into why our on cluster logging tools weren’t ingesting anything would have been nice.
What support did you contact if you weren't paying? Support isn't perfect but I find it hard to believe you got no response at all. Staff and community is also active on the gce-discussion [1] group and StackOverflow [2].
Based on how Google is quick to mothball products, I can't honestly trust the GCP for production-grade services.
Unfortunately, there's not much the Cloud team at Google can do. Google as a whole needs to demonstrate long-term support in a superior fashion before I can trust their products for building my products.
Why is this getting downvoted? Provide some input on it, has Google suddenly stopped killing projects that are in their infancy? Why should I entrust my platform to that behavior? Anything instead of just mass downvoting.
You're getting downvoted because you don't base your statement on anything. Google sometimes cancels Early Access or Alpha stage products. Beta products may change before they are made Generally Available (GA), but I cannot personally remember any beta product being significantly changed or cancelled before GA. Any GA product will be stable for quite some time, you will get at least a 1 year notice before they deprecate anything, as can be read from their terms: https://cloud.google.com/terms/ section 7.
But I understand hat some customers require enterprisey products that stay exactly the same for at least 10 years, and a 1 year notice might not be enough to make adjustments.
I don't care to "hunt" for shills anyway, but it's real frustrating to make a comment against Google, get hit for it, and the best "response" is a wall of text just making it "OK" for them to kill products off left and right early. As if this company-wide behavior should just be discounted because "Google".
The new CEO is the former President of Product Development at Oracle...I wanna puke. Oracle has some of the most hated products in the industry. I hope he's able to adapt to Google's style and not the other way around.
He's going to be used as the heavy-hitter salesman that goes around and says "I see you're using Oracle, I know Oracle, trust me when I say GCP will meet your needs" and he'll have tremendous credibility saying that.
Not to be rude, but yes you are totally wrong here. Meeting with customers is 50% or more of the time commitment for a world-class enterprise-sales-company CEO.
Other companies are huge contracts, and simply won't sign on without that level of dedication. They demand it, and the CEO's give it.
It's just as important at a bigger company. CEOs end up traveling more as the company gets larger, mainly to meet customers, and it's especially critical if the customers are also big companies.
CEOs and CTOs are often involved when working with a very large business. CTOs often end up giving presentations to the top brass about security and technological advantage. If you are high enough to have major influence on the product, you will end up doing sales.
That's not what someone in his role is expected to be doing. He may support, or defund, certain initiatives, but he won't be the "originator" of anything major.
The better question is, "But what type of people will he gravitate towards hiring to develop some products?"
Except he's a deeply technical product guy with 20 years experience managing the development of the worlds most profitable database.
Its funny that Hacker News hates on Oracle, but their database is far ahead of the competition in a highly competitive market. Say what you want about their other products (that are indeed terrible), but that core database gets them billions in revenue for a reason.
A good salesman knows keeping credibility is more important than keeping a job. I quit my position as a sales engineer when the company asked me to lie about the product's capabilities.
This isn't the first time Google has gone outside the company for a top leadership role. Vic Gundotra was another one--and that didn't work out too well.
But this is happening with increasing frequency even for mid-level leadership positions, and it is showing more and more in the culture.
Is there a reason to why google is doing this? From what I get from my friends Google has issues with employee growth and it's pretty easy to stagnate. If google keeps going outside for leadership roles then is it overlooking capable internal employees?
It's because Google's culture is engineering focussed while pretending to be product focussed. As a result, Google breeds great engineers but not great product people. Also, their promotion process shamelessly promotes peer cannibalization. So, its easy for naive people to stagnate.
It means that making your peers looks bad is as important as making yourself look good. There's only so many spots for promotion, so employees have to stab each other in the back in order to look impressive enough to be promoted.
i’ve only noticed at least teams/products in google that are more stagnant or don’t have as much work to do. therefore they compete and are territorial with the work. other products that are growing there is plenty for everyone
I think Google just can't find people internally who want to do it.
I know of two groups (one of maybe 10, and another maybe 30 engineers) that sat essentially leaderless for several months each before Google hired a manager from outside the company. Neither one of these, in my opinion, have a good handle on Google culture.
There is no real reward for becoming a manager (no more pay, no more level, no more nothing), but a lot of hassle, so very few engineers want to do it.
At least that is what I see from my limited vantage point.
It's also hard for those Googlers who want to do it to be given the chance, unless already senior enough on the technical ladder to correspond to the top 20% of technical people at Google (a far higher bar than the technical competence required of a manager).
Source: personal experience circa 2013-2015, when I was there and trying to transition from individual contributor and tech lead to first-time people manager. Of course I wasn't considering executive roles like this article is about.
Companies typically hire outside people for leadership roles when they want to change their culture or gain some new industry connections. Promoting someone from within will reinforce the existing culture, for better or worse.
They are good in their space, but aren't as good as companies that have more experience.
IBM, CA, etc can sell pure garbage because their sales folks understand how the game is played, who to please, etc. Amazon won't even do simple stuff like relatively minor redline changes to contract terms.
Also, by "good" I mean maximize revenue/sales/renewals. Not specifically delivering something good in the end.
At that level, he just isn't the guy making the decision to add a boolean type to the DB server or not. According to wikipedia, Oracle has 3000 products, each with their own management.
The role is organisation and team psychologist more than API design. If you want to rate his performance, the products are a worse indicator than, say, how well-adjusted and mentally stable his teams were...yeah, you're right. GCP is fucked.
Although I'm not so sure what Google's style is, it's been changing so much in the past few years, and a lot of their slimy stuff has been exposed too.
Gonna be totally honest: From the headline and Google’s previous behavior, my first thought was “Huh, I didn’t expect them to shut GCP down already...”
(To be clear, they are not shutting down GCP, this is just a leadership transition)
I think one of the big issues that Google is facing with GCP is exactly the perception they have in the market with respect to product shutdowns.
I could never see myself making the decision to go with Google over Amazon or Azure, simply because I don't feel confident that the product won't be shutdown with minimal warning at any given time.
It is strange that smart people like this are shooting themselves in the foot in this way.
Yes you can save money by killing a product, but I would argue the second order effects are more important. People lose the faith to bet on your products.
Having had to scramble - twice - when google cut products under me, I would never make a big bet based on an assumption that they will keep faith after I commit to their products.
Maybe they should read H Edwards Deming e.g. "maintain consistency of purpose".
This is like those database abstractions that get made under the premise that someday at some point you'll switch DB's. But yet, 10 years later you are still chugging along on the same database you started with. You would have been far better off to just completely "move in" with the DB you decided to use in the first place and leverage all it has to offer.
The problem with these kinds of abstractions is in order to avoid "vendor lock in" they are forced into a "lowest common denominator" deal where you don't get to leverage any vendor-specific awesomeness. For example, in DB land you would avoid certain kinds of database fields (Postgres's structured datatypes like JSON, for example). Instead of leveraging the DB to do certain kinds of JSON queries you now have to write the infrastructure your self "just in case we switch DB's".
The irony is, in my experience to avoid the dreaded vendor lock in "problem", teams and organizations wind up locking themselves into their own homebrew nightmare that does nothing but recreate (often poorly) what they could have gotten by allowing themselves to get locked in to a third party. It goes without saying that in almost all cases, this homebrew nightmare has zero to do with the actual value generating functions of the business. Storing and querying on structured JSON in a database has nothing to do with the core values of 99.999% of all organizations--these things should be locked into third parties.
My point is, most teams should embrace all their cloud providers can offer. Settling for the lowest common denominator shared between cloud providers leaves money on the table and runs a very high risk of what is, in my experience, a worse form of vendor lock in--getting locked into your own stinking pile of rubble.
I could go on and on about the "vendor lock in" boogyman--it is one of the biggest, most annoying anti-patterns I see. It creates huge political and technical hurdles as the business grows. I believe it stems from a poor understanding of actual business risks, not understanding the true lifetime costs of inhouse engineering projects, resume-driven design, and a very healthy dose of premature optimization. (also toss in some Stallmanesque mistrust of proprietary systems for good measure)
+1 to the above... There's HUGE costs in trying to create abstraction... Generally layers of crap code that is all repetitive and adds no value, making it so much more costly to maintain even. You're often better off writing a second version and transition scripts than trying to maintain an abstracted code base or application platform.
Though a lot of the docker tooling, k8s and other options do make it quite a bit easier as an option of you want to self-host. But that has other issues, advantages and disadvantages.
I will say that the majority of Rails apps can just use the Rails abstractions and they sit reasonably well on top of whatever database you choose to use. Rails 5 maps just fine to most common database types and patterns.
On the other hand, in the past I’ve used Fog to abstract our cloud provider and that was nothing but pain. SQL is well-specced with some vendor-specific extensions, while there really isn’t any actual common denominator for cloud services.
Have you been using the same cloud abstraction tools I have? Terraform and Kubernetes all require more effort in dealing with cloud providers, not less.
I bet Google would gladly undo the Reader decision now, and even pay whatever they saved in costs over the last ten years, if only it would spare them from the endless repetition of this cliché.
FWIW I'm pretty sure Google projects have at least as good a life expectancy as startups do.
Even if they hadn't shut down Reader, there's still Hangouts (v1), Talk, Chat, Messenger, Wave, Buzz, Plus, Hangouts (v2), Orkut, Helpouts, Inbox, Knol, Video, Nexus Q, Spaces... The list just goes on and on and on.
How can it be worse than a leader from VMWare? Have you ever seen VMWare's cloud product? It literally has little pictures of pretend computers that you click the reset buttons of with your mouse.
Am I the only one who has had much better experience with kops or just straitup kubeadm then aks/eks/gke? Stock k8s in my opinion is not hard to bootstrap.
They are fake internet points and don't matter. Making comments about downvotes usually means more downvotes.
While votes are usually reserved for content quality, HN also finds them acceptable for disagreement in opinion, so most likely people disagree that stock K8S is easier to setup than the managed offerings.
Based on my own experience as the CTO: After being an Azure shop for a year we've migrated ~50-100 VMs to GCP and I love the GCP products.
GCP is:
- Simpler to use
- More tailored to people with Linux environment
- Leader in K8S
- Has good support
- So much cheaper (in our case we saved ~60%)
- Has great UI and understandable primitives.
My only pet peeve is the fact that exporing your spend is practically impossible unless you're a BQ guy that can work directly with report exports.
PS: We're building our future infra on K8S to allow us to migrate more easily to a different could if something goes awry with GCP, I really hope there won't be a need to migrate back to Azure and its arcane and high pricing, strange UI, worse tooling...