Help get this topic noticed by sharing it on Twitter, Facebook, or email.

Scheduled Downtime for Springpad - 11/18/11

Springpadit.com will be unavailable this Friday night 11/18 for approximately three hours beginning at 11pm (ET).

We will be performing for maintenance and upgrades to enhance Springpad’s speed and reliability.

While online access to Springpad will be unavailable, any data that you have already saved will be accessible via the Springpad Chrome App (in offline mode), and via the iPhone, iPad and Android apps.

More details: http://blog.springpadit.com/2011/11/s...

Sorry for the inconvenience!

----
UPDATE 11/21:
Hey guys! Just wanted to follow up with more details about the downtime over the weekend...

On Friday night, we upgraded the Springpad database to the latest version of Cassandra (http://cassandra.apache.org/ ). We were a few versions behind & felt that the new Cassandra offered some great performance and reliability improvements. We initially planned for 3 hours of downtime on Friday night to do the upgrade.

One of the steps of the upgrade is to update the data files (all of your data). We had chosen to do this in an incremental way, planning to do the upgrade of all of the data files over the next week or so. But once we began that update, we saw a huge performance penalty for using that option - it basically made Springpad unusably slow. So, on Saturday afternoon, we decided to bite the bullet and perform the full upgrade all at once. This, unfortunately, meant another 3-5 hours of unexpected downtime.

While we are sorry to have disappointed you with the unexpected downtime, we do feel that the upgrade will greatly enhance Springpad performance over the next couple of months. Plus, we have some great new features coming soon, and we needed the upgrade to support them.

Of course, our goal is 100% uptime, and we realize that anything short of that is unacceptable. To avoid something like this from happening again, we're improving our already-exhaustive testing process to better simulate user load. Hopefully this will help us to better predict downtime situations like this, or (even beter) avoid them altogether.
Reply