Comments on: Backblaze open sources 135TB storage architecture

By: Faiyaz Ahmed

Faiyaz Ahmed — Thu, 21 Jul 2011 18:13:03 +0000

Good architecture, and suits their needs. But I wonder about fault tolerance. You’re at the mercy of md to fix your raid group, provided you have enough spares to rebuild the volume. The buy-in for the more expensive alternative is the ability for fault containment, migration, and expansion. Particularly, the netapp and emc boxes- you can expand the volume without downtime and without compromising redundancy. But, then again, that isn’t blackblaze’s usecase or target industry.

By: Don Honabach

Don Honabach — Wed, 20 Jul 2011 19:59:46 +0000

True — maybe I missed it though — did they ever claim that they were designing high performance storage systems? I think for their core mission, this solution is great and a perfect example of how one can use commodity parts to solve problems. For their needs it makes a lot of sense.

By: Don Honabach

Don Honabach — Wed, 20 Jul 2011 19:48:46 +0000

These guys are great — they helped with Q&A and I was actually one of the first folks to build one of their earlier models outside their company (http://extrememediaservers.blogspot.com/).

Not many companies are this forward thinking!

By: AffirmedSystems

AffirmedSystems — Wed, 20 Jul 2011 16:13:51 +0000

This is a low performance architecture on high latency SATA disks… anyone telling you otherwise is a wrong… try running real-time applications on one of these clouds – you’ll want to jump out of an airplane with frustration….