A couple of service updates
December 22, 2010 | David HenzelHi Everyone! It's been a crazy December. We hope you're all excited for the Holidays and New Year. You can bet we are! There are a couple of service updates with the MaxCDN/NetDNA control panel we wanted to quickly address: 1) PHP/Zend Errors Starting Sunday @ around 4:30PM PST, some PHP/Zend Errors were showing for some users in the control panel randomly. It took us a while to track down the cause - we had recently migrated to AmazonRDS and the instance was rebooting from mySQL crashing around once an hour. We tracked it down to a coding error with how we were storing sessions for our API and it should be running great now and scale very gracefully. 2) Re-Indexing Popular Files Report Based on feedback, we've decided to re-index the Popular Files report. We originally didn't design this well to accommodate date range searching for the tens of billions of requests per month. Rest assured we're on top of it and the Popular Files report will return to the control panel shortly. We're also working on a release of our control panel with lots of cool reports and widgets based on what we here from you! 3) New iterations of Pull Zones 2.0, Live Streaming 2.0 and Push Zones 2.0 We've quietly been working on new iterations of what we call "CORE" internally. CORE is the software we've built to service Pull Zones, Live Streaming and Push Zones on our network. If you'd like to migrate to version 2.0 of our zone, feel free to reach out to me and we'll get you on the beta list. Pull Zones 2.0 (Beta - Available now):
- HTTP 1.1 Compliant on the origin/backend -uses keep-alive and honors Gzip headers. (We've always been HTTP 1.1 compliant to the browser)
- Honoring DNS lookups for origins - we use a super fast DNS resolver every 5 minutes to make switching your origin host pain free (no more perminant IP caching). It also works great with Amazon S3's DNS redundancy.
- New Caching Algorithm - Not all content is equal - so we now use a caching algorithm that's based on "hot" content rather than just space. It's more efficient in achieving high cache hit percentages and keeping the hottest content in all the right places.
- Bigger SSD arrays- Pull zones are already 100% SSD based. We've recently tweaked a high performance, low power super cluster which allows us to be ridiculously efficient on bandwidth per server and bandwidth per watt of datacenter power. We have an even more high I/O iteration of the server platform going into production in January. What does this mean for you? We're efficient and can provide value and speed for you.
- New Acceleration Stack - we're going to be making a public announcement of a partnership in the coming weeks, and it gives us access to a great stack for accelerating TCP Slowstart and TCP Congestion. What this does is extend the reach and effectiveness of our PoPs. Anything over 10ms in latency will see ~20% improvement and the larger the latency and the bigger the file, the more improvement (upto ~200%). The new stack also provides very fast and consistent delivery to highly latent environments like mobile devices, WiFi and even airplanes wireless.
- Faster Replication - We've come up with faster replication techniques so you don't have to wait.
- New Acceleration Stack - (See the above description from Pull Zones)
- Dedicated Storage - we were previously using NFS with attached storage and have itterated on a new direct-in-server storage that's fast and efficient (and good on datacenter power!)
- Universally Good Publishing
- Stream Repeating
- Less Jitter