Google Compute Engine vs. PyTorch on AWS

It is excellent if you have any workloads that need raw computing or plan to have any state-full services running in your environment like DBs (for which you don't want to use Managed services), cache, etc. It also gives you complete control over which versions of software, OS, etc., you need, and thus, you can build anything and deploy it on GCE.

Incentivized

• A simple web-based interface that is a breeze to train new engineers to use. Our experienced engineers never have trouble finding or doing anything on GCE.
• Sustained use and Committed use discounts mean we get top-tier VMs for an incredibly competitive price.
• Wonderful identity and access management that gives us peace-of-mind when granting access to machines to contractors and other 3rd parties.
• Fast VMs, lastest in hardware, and enough RAM to power even the hungriest of our services.

Incentivized

• The L7 load balancer can be difficult to get set up. It's limited in its functionality, especially with the container engine.
• It's hard to find certain objects on the web console. Often times the things I need to get to are buried in advanced menus.
• Google's decision to only support MySQL on their relational DB service means that I have to manage Postgres instances in Compute on my own, managing everything from storage to backups.

Incentivized

Its pretty good, easy and good performance. Also, interface is very good for starters compared to competitors. Infra as Code (IaC) using Terraform even added easiness for creation, management and deletion of compute Virtual Machines (VM). Overall, very good and very easy cloud based compute platform which simplified infrastructure, very much recommend.

Incentivized

Having interacted with several cloud services, GCE stands out to me as more usable than most. The naming and locating of features is a little more intuitive than most I've interacted with, and hinting is also quite helpful. Getting staff up to speed has proven to be overall less painful than others.

Incentivized

Google Compute Engine works well for cloud project with lesser geographical audience. It sometimes gives error while everything is set up perfectly. We also keep on check any updates available because that's one reason of site getting down. Google Compute Engine is ultimately a top solution to build an app and publish it online within a few minutes

Incentivized

The raw computer power is excellent; our applications feel snappy, pages load almos instantly for our customers and so on. The primary reason it is not a perfect 10 is that the native tools for monitoring individual VM performance can be complex, making it challenging to easily diagnose specific resource bottlenecks without significant configuration

Incentivized

• The documentation needs to be better for intermediate users - There are first steps that one can easily follow, but after that, the documentation is often spotty or not in a form where one can follow the steps and accomplish the task. Also, the documentation and the product often go out of sync, where the commands from the documentation do not work with the current version of the product.
• Google support was great and their presence on site was very helpful in dealing with various issues.

Incentivized

When configuring Amazon ECS, it is a bit confusing as you are not able to find the actual issue. You need to enable Additional AppInsights to get detailed level info, which is not a concern when configuring on the Instance Level. Moreover, Azure VM does not provide an in-browser option; instead, it is Azure Bastion, but for that, you have to enable a dedicated subnet, which is a bit unnecessary.

Incentivized

It works really well with other Google Cloud services, making it easy to build scalable solutions across different teams and locations.

Incentivized

• Scalability means flexibility and less upfront costs
• Can become expensive when hard set compute requirements are clear, but things like Spot VMs can help here too, or just having your own infrastructure and scaling up with Google. This is for more advanced cases though
• Ramp up time is long, but after that it is quick to do many things and ROI is awesome

Incentivized