MediaTemple struggles with outages

Utility computing only works when uptime is near 100 percent. Just ask Media Temple, which is struggling with its uptime of late.

The outages of MediaTemple's GridServer are starting to balloon. The company was put on the radar by TechCrunch and highlighted on ZDNet and elsewhere. The problem today: Early adopters are getting irate.

For now, MediaTemple is struggling to recover, but has at least been alerting its customers about its progress.

Among the details:

Nov. 24:

At approximately 7AM this morning, customers on GRID.Cluster.1 experienced a brief interruption in service with the http segment of the cluster that lasted approximately 20 minutes. This incident was due to some aggressive performance tweaks that were set on Wednesday. We have since adjusted this setting and will be investigating this issue further.

Nov. 28:

Customers with (gs) Grid-Servers located on GRID.Cluster1 may be experiencing issues with accessing their websites, mail, FTP. It appears that the Titan storage segment powered by BlueArc (http://www.bluearc.com) was disrupted due to a bug in the system. (mt) Media Temple systems engineers are currently working with our vendor to apply a firmware update and resolve this issue as quickly as possible.

Later on Nov. 28:

After extensive troubleshooting and testing, BlueArc engineers and (mt) Media Temple engineers discovered the main issue which was causing performance problems within the storage segment. Together, the system engineers made adjustments both to the BlueArc Titan system and the GRID email segment layer which have alleviated the performance related problems experienced after the latest firmware update was applied. There remains an unresolved issue with exorbitant file locking which has not yet shown to cause availability problems, however warrants us to keep this issue open for another 48 hours as a safety measure. Also, as part of this evening's configuration adjustments, (mt) Media Temple will be conducting a very short window of maintenance (5-10 minutes) to the storage segment between 2:00AM and 4:00AM which will result in brief unavailability of services in GRID.Cluster1. After this maintenance action is concluded, (mt) Media Temple will increase the monitoring of this segment watching carefully for any reoccurrence of the symptoms which caused today's issues.

Further updates will be provided as results come in from the extended monitoring. We thank you again for your patience and apologize for any inconvenience.