(gs) Storage Updates and More…
In this update we have a progress report on our transition to a segment-based storage model using Sun hardware, some good news about our cluster migration tool, and we’re announcing the start of our MySQL 5 Upgrade BETA program. We also have an update on the anti-spam and blocklisting front, a followup to our transparency initiative, and some stuff about our future capacity expansion plans. In addition, we’re throwing in an interesting discussion from the (mt) User Forums that our (gs) customers might find interesting.
Cluster Upgrade Progress
Our Gen 2 storage segments are working very well. Vendor procurement and upgrade progress continues to be smooth. Customers new and old are confirming daily their service levels are solid, even as migrations continue to run in the background. We are committed to keeping people updated on progress with these migrations, and there are currently 18,000 sites left in the queue. For those anxious to cut in line, as promised, we have a new tool.
Cluster Auto-Mover
As promised 30 days ago, we have developed a cluster-to-cluster movement system. Internally the system is code named Fiji Agent, and it is in testing now. Fiji Agent will allow customers to have more fluidity within the GRID overall. For example; today it will be used to move customers from 1st-Gen to 2nd-Gen cluster technology. In the future, we will extend Fiji Agent to work with (dv) and our forthcoming tech. Fiji Agent could be used to try “beta clusters” for example, or change data center locations completely.
We estimate that within 7-14 days the tool will be available to customers. Look to this blog for forthcoming announcements on this progress, as well as instructions for getting your own sites migrated.
More Grid Improvements
As we’ve continued work on storage stability and migration tools, other notable progress to the (gs) system has been made in parallel:
System-Wide MySQL 5 Upgrader BETA
Thank you everyone for being patient with this one. After an extensive development process and hardware deployment initiative, we are now activating select customers in the (mt) Labs – MySQL 5 Upgrade BETA . Those users will be getting an email very soon.
Highlights:
- Auto-Upgrade to MySQL 5 using newly designed tools and interface in the AccountCenter
- Early access to MySQL 5 for Clusters 1-3
- (mt) User Forums – ‘Top Feedback’ program
To ensure a short and focused BETA period we have ended new sign-ups. We’ll be releasing this system to the general public within the next 30-60 days. **Note: All users who purchased a (gs) account after 1/2009 already have access to the MySQL5 system.
New Anti-Spam Weaponry
We had some widespread blocklisting issues on 3 of our clusters throughout the month of March. This was largely due to new spamming methods abusers were using that had not been seen before. We caught on to it pretty quickly and have introduced several changes to combat this behavior. We also have been persistent in our efforts of clearing up all clusters from any known RBLs. Avid Twitter’ers may recall us contacting Comcast over the popular social network to resolve a blocklisting issue (thanks @comcastbill, @comcastcares!).
Equally important, Engineers have designed new detection code and mail flow controls that automate many tasks for our email administrators in this ongoing battle. We are already seeing big improvements behind the scenes. Thank you to all of the customers who have given us ticket feedback on this subject.
One change worth noting – We will start blocking outgoing mail over port 25 (sending through your email app will still work fine). This is an email sending method used by less than 0.25% of our entire customer base. In most cases, usage statistics indicate these are mostly spamming attempts. We will make a separate announcement with more details when this is in place.
Improved Transparency and Incident Handling
In our Anatomy of Storage on the GRID post, we promised to improve our transparency and coverage of System Incidents. Since that post we have successfully provided comprehensive public coverage on roughly 15 GRID-related incidents. Internally we have made numerous improvements to the way we work and communicate.
A new enhancement recently made available is targeted System Incident notifications from within the AccountCenter. Customers affected by a currently-occurring incident are notified immediately upon logging in to the AC, with the affected hosting service specified, and are guided to the public System Status page for details. Features will continue to be added to this tool including enhanced incident notification methods.
We understand that even the best System Incident coverage possible is secondary to having a consistent and stable hosting experience. That is definitely our top priority, but we’re glad that our reporting is helpful to so many customers.
Infrastructure Expansion
We have received new hardware that we will be using to expand our capacity and improve performance for the GRID clusters. These are planned for production deployment within the next 15-30 days. We have also deployed various bug fixes and optimizations that will improve everyone’s hosting experience. Cluster 5 progress is coming along nicely, as well.
From the (mt) User Forums
Customers in our forums have gotten together to share benchmark statistics. In this interesting discussion, regular users compare performance benchmarks between sites on the original (gs) architecture, and the new generation currently in use on our later clusters. It’s definitely a worthwhile read.
Performance Benchmark Comparison – 1st Gen VS 2nd Gen: Jump to the Thread
If you’re not yet registered in the forums, see this KB article for super simple registration steps.
More To Come…
The future is very exciting and more improvements are on the way. We plan on making these sorts of update posts a regular thing, so stay tuned to the blog!
