How Google runs its storage and more from the storage blogs

Google has posted a paper that outlines its storage architecture, and that’s the choicest morsel in a month when the usual blogosphere protagonists have gone easy on each other.

How does Google do it? How does it store all those zillions of emails and documents?

Robin Harris of StorageMojo has found out, pointing out this paper entitled Megastore: Providing Scalable, Highly Available Storage for Interactive Services that explains just how Google, to borrow Robin’s preamble “handles over 3 billion writes and 20 billion reads daily on almost 8 PB of primary data across many global data centers.”

We’re going to read the paper as soon as we finish writing this, which is a little less spicy than it usually is because storage bloggers are mostly ignoring one another at present.

Storage Anarchist makes some comparisons between tiering architectures and is a bit mean to 3Par (whose “ chunklet architecture requires so many SSDs to balance the load across their nodes as to be cost-prohibitive”) and IBM happily contests the way ‘Zilla roars about cloud models. Technical Deep Dive seems to be a bit mean to IBM and the VCE Gang, but that’s about it on the competitive narkiness side.

Instead of the usual spite, this month we have an outbreak of lists. StorageIO has a fine list of jargon acronyms, as does Desinoscloud. Preston de Guise has five thoughtful questions for Drobo and Storage Buddhist has a wishlist for the ultimate array.

A couple of posts riff on lockin, namely DataCore’s Australian team and Chuck Hollis.

Speaking of Australia, there’s some interesting marketing going on over at NetApp, which has made a video about a win at local financial services group Suncorp.

It’s also made a video with Dave Hitz and an accomplice talking about a whole bunch of stuff Suncorp hints at but doesn’t say straight out in the final cut. Your correspondent has never been a fan of the journos talking to journos sub-genre of journalism, so is not entirely sure what to make of the executive asking marketing person how great the marketing is sub-genre of social marketing. Also on NetApp, StorageBod looks at its decision to resell Stornext and Eigenmagic explains thin provisioning with the big blue staple.

While we’re on case studies, here’s one about EMC in Alaska. At this point I’d usually go for a cheap Sarah Pailin gag, but Donald Trump has made her kind of obsolete.

There’s a bit of chat around about AWS’ failure. That one’s from The Storage Architect and you can google the rest because they’re not super-interesting, just like this Dell post on its storage plans given that what we all really want to know is how it’s love/hate relationship with EMC will play out. Of more interest is Claus Mikkelsen on HDS’ UK Geek Day which also scored digital ink from the Architect.

This one, however, is interesting. Canned Platypus coins a great term: the Entreposeur. Have a read if you’re tired of people who endlessly talk about the great startup they’re working on.

Perhaps oddly, we could only find one post on recent disk mergers, from Steve Foskett, who also came up with this interesting piece about Thunderbolt, the new super-fast fibre-channel derivative that Intel and Apple have slapped into newer Macs and which he says may herald a new mode of Mac-to-SAN connections.

Storage Switzerland has a look at the future of tape and ESX Virtualization points an ESX host at a FreeNAS and notes the results. Ray on Storage notes a CDMI plugfest.

Lastly, some mixed news. Farley has retired StorageRap but promises “a new multi-media oriented storage blog.” Farley has a great face for podcasts, as we say in the trade, so the new venture should be interesting.

Big cheers for Chad Sakac’s wife Susan too, as he blogs she ran the Boston Marathon in just under four hours. Your correspondent would struggle to double that distance in the same time on a bicycle, so kudos to Susan!

Read more on Storage management and strategy

CIO
Security
Networking
Data Center
Data Management
Close