Maximizing Cloud SQL database availability

Key Takeaways

[Music] hi everyone I'm here at next24 with Rahul and we're going to talk about availability in Cloud SQL and you're giving a talk yes I am going to talk about how our customers can maximize the availability of their Cloud SQL instances wonderful okay and so you mentioned maximize availability and downtime so what are some of the most common causes of downtime in Cloud SQL yeah so uh again Cloud SQL is a managed database service we offer a 99.99% uptime SLA so we hold a very high bar for ensuring that your instance is available but typically what we see is that a lot of day two operations like instance maintenance customer when customers needs to scale up their instances that typically results in a downtime and then of course there is the you know issues with the infrastructure or the D itself that can also cause some downtime right right makes sense and so I recently heard a little birdie told me that we're at what's considered no downtime now less than 1 second amazing how how can gen apps really benefit from this new capability of almost no downtime or no down time actually yeah so yeah uh so uh last year when we introduced Enterprise Plus for cloud SQL we introduced what we call as a near zero downtime plan maintenance and at that time we said that we are reducing the downtime to less than 10 seconds and I'm very happy to say that we have made so many improvements since then that we have been able to reduce that downtime now to less than 1 second and we technically calling it subc downtime but you know you can call it you can call it zero downtime if you really want right yes um and then uh to your question about how gen applications can benefit you know a lot of our customers are using generative AI to build applications for critical Industries like healthcare where the only acceptable downtime is zero so Cloud SQL can really help them you know where they can do their Vector search and Vector embeddings within say cloud SQL for post SQL using PG vector and they can do that confidently knowing that when maintenance happens they won't even see it happening amazing so going off of that a little bit can you give me a little bit of behind the scenes of how are we able to achieve this subc downtime so the way that we do it is uh what we are what we're doing in under the hood is we are creating what we call as a shadow replica and we make sure that the replica is fully caught up with the primary instance and as soon as that happens we just do a quick cut over and you know the we start redirecting the traffic to the new primary instance and to the application and to the user it seems like nothing really happened there was a quick blip and the application is back talking to the database again wow that's so cool um in terms of making sure that we follow best practices could you highlight some of those that are often maybe overlooked but are really significant for high availability yeah so uh you know availability within a region is really well understood at this point and a lot of our customers take the advantage of the cloudsql high availability solution where they just check a box and they get the protection from any kind of a failure within a Zone but what we have seen is customers still struggle with you know setting themselves up for Disaster Recovery where they are ensuring that if uh something happens to a region that they cloudsql instance can they can they're able to start it quickly in the other region and for that what we introduced in public preview this week is something that we are calling as advanced Disaster Recovery with Advanced Disaster Recovery what happens is is you can create a cross region replica and if there is a disaster if there is an issue with the primary region you can quickly go and promote your cross region replica and then when the original region comes back up we Cloud SQL automatically take care of recreating your old instances and then when you need to go back to your old region because that's where your users are we offer you a simple operation called a switch over where you can go back to your original region with zero data L wow okay that is amazing um where can people find additional resources on all of this awesome information so the cloud SQL documentation is perhaps the best place to find all of these details where you know where we talk about how you can uh what how these features work and how you can leverage them we have other resources in the making but right now docs is the best place for you to go wonderful thank you so much for being here with us today Rahul and have a great of the time at the conference thank you it was great to be here and I'll see you all [Music] soon