Amazon Cloud is down? Data Center Strategies.

[Samantha Morris]

Loved this tweet this morning into all the talk around the Amazon Cloud going offline:

@ruv: Get over it. Backup, have a disaster recovery & risk mitigation plan. Use more than one DC provider. <-#AWS cc: @racheldines

Couldn't agree more...but Reddit, Foursquare AND Quora, along with a host of other websites are all experiencing outages. These companies are all startups, very popular startups but shouldn't they know better than to use a single-vendor strategy for their data centers?

What's your take on this? Is anyone affected by this today? What kind of strategies does your organization have in place to prevent situations like this from occurring?

If your data center goes down, where else do your backed up files go to?

If you are (or hypothetically) using Amazon EC2, supposed disaster was to strike your organization or client site today, how would you recover the data to ensure business continuity?

[Ashar Baig]

Large IaaS providers like Amazon do not provide SLAs around Disaster Recovery (DR) and Business Continuity ((BC). The Amazon outage reminds us that we must ask our Cloud Service Provider (SP) to provide SLAs around data protection to ensure BC. These SLAs should have non complaice clauses (teeth in the contracts) built into the agreements.

[chapa]

This is really very simple. If cloud will be an extension or a replacement of your onsite operations, plan accordingly. Just because its in the cloud doesn't me it is protected.

I just wrote a blog on it with my perspective: "Just Thinking Out Cloud" http://www.navigatingthebarscene.com...ing-out-cloud/

[rahrenstorff]

It's not even so much a single vendor problem; those startups chose a single data center without implementing contingency services offered by that same vendor. I presume it was done to keep costs down and they probably didn't anticipate the mighty Amazon would experience an extended failure.

The interesting data point for me is; did the Amazon services provide a measurably better ROI with an equivalent service availability metric (compared to internally deployed IT systems within a single data center).

As a provider of cloud backup services powered by Asigra, we operate two geographically dispersed data centers to avoid single site failures.

Rodd Ahrenstorff

[smokey]

One obstacle with multi-vendor type solutions is that vendors hate playing nice with each other. Easy migration to another vendor is not in their interest.

Also, for whatever it is worth, the amazon outage was not isolated to a single data center. The problem was misconfigured site to site replication so it impacted the entire infrastructure at all locations for the EC2 service.

http://www.pcworld.com/businesscente..._question.html

[rahrenstorff]

Smokey is quite right and worse yet, permanent data loss occurred for some customers: http://technolog.msnbc.msn.com/_news...customers-data

[Handy_Backup]

Feels a bit strange to observe vendors treating one another that way. On the one hand it is totally understandable, as they dream of dominance over customers' data. But on the other hand the win-win cooperation, when you can't at once deliver the exact product or service a customer wishes, so you redirect the customer to a proper company, has been resulting in notable profits for IT conpanies practicing it lately. I believe there exists an akin strategy for vendors which might both support customer data securedness as outages happen and be at the same time profitable for the vendors. Hope it's the matter of time.

Aleks Dolgushev