Feed: Planet PostgreSQL.
It goes without saying that this service is a direct competitor to Amazon’s RDS and understandably there will be comparisons going forward, both in this post as well as anywhere else Cloud SQL gets reviewed. With that said, lets dive in!
Account for Getting started: Currently, Google Cloud Platform is providing $300.00 in credit for maximum of one year. That is great, especially if all you need the platform account for is to try out Cloud SQL for Postgres. Setting up the account is super fast since like most things google, for better or worse, the account you’re signed in as is used to activate the cloud platform service for you and you’re in your console in a matter of seconds. Amazon too has a generous plan for beginners to get into their AWS world.
Documentation: Cloud SQL’s documentation is very clear and concise. Also appreciate the direct links to sections of the console within the tutorial steps. Very handy and makes it immensely easier to find your way if, like me, you have never worked on Google’s Cloud Platform before.
Interface: I found Google’s instance creation interface to be much cleaner than Amazon’s. While, to create an RDS instance on Amazon, you’re guided through multiple pages starting with the basic settings and progressing to the advanced ones, instance creation in google is all done on a single page hence making the process seem faster and simpler, also allowing to review as you choose new settings. This may change though, as Cloud SQL ads more features and options.
Security: While Google’s instance creation page is simpler, I did not see any security group/VPC/Subnet/VPN related information I was required to sign off on before my instance was created, unlike RDS, where the minimum barrier to create an instance is to review these settings and hence is much higher than the minimum barrier in Cloud SQL. My guess is google prioritized speed and simplicity of instance creation over security, which might be good to get in customers but not as good in the long run.
Monitoring: Basic monitoring in both RDS and Cloud SQL are enabled by default. RDS instance creation provides the option to choose enhanced monitoring at time of creation, whereas Cloud SQL does not.
Logging: Logging in Cloud SQL seems to be better than RDS. There are options for extracting log messages based on logging levels. As far as I know there is no such options RDS
Memory: Google’s memory options only span from 3.75 to 6.5GB, while RDS’s is that from 1 to 244GB. That is a huge difference between the two offerings, but I suppose being in Beta, Cloud SQL’s actual services may change overtime.
Storage: This ranges from 10 – 10230GB on Cloud SQL and from 5 – 6144GB. Unlike memory, Cloud SQL is the one with more options when it comes to storage.
Customizable Configuration: Not too impressed with Google’s Cloud SQL Flags. It is very limited at the moment. RDS itself only provides a portion of the configuration parameters to be customizable, but compared to google’s flags it is a lot better. Currently, only autovacuum settings and default_statistics_target
Pricing: Prices for Google Cloud SQL for Postgres and Amazon Postgres RDS look quite comparable at first glance, but on a closer look I found that Google’s pricing is a bit higher than Amazon’s. Whether the cost difference is worth it will be clearer in the next few months when the final service is released and more details about performance come out.
At first glance, I am impressed with Google Cloud SQL, mainly because of the ease and simplicity of setting up a working cluster, and the monitoring and logging interface. The one thing I did not like, especially when compared to RDS is the lack of advanced network settings such as setting up a VPC and security groups in general. These may be accessible elsewhere, but most beginners creating an instance may not go through the trouble of searching for these. I think Amazon does the right thing here by forcing the user to review these settings, even if it comes at the cost of simplicity and speed, even with a light headache at times. As for the extremely limited set of customizable flags for server configuration parameters, I am pretty sure it will change and grow as more people will start using this service.