Give Rackspace Cloud credit: As part of a system upgrade, RackSpace Cloud is proactively telling customers that their sites may be offline for as long as 90 minutes on March 6. As a Rackspace Cloud customer myself, I appreciate the heads up. Still, I’ve got key concerns and I have to ask an obvious question…
… aren’t cloud platforms designed to be redundant, so that no customers go dark during system maintenance? I’m not the most tech-savvy guy. But it seems rather wild to think that some RackSpace Cloud customers (including MSPmentor and its sister sites) could be dark for as much as 90 minutes on March 6.
According to an email I received from RackSpace:
“The Rackspace Cloud’s system engineering team will be performing several upgrades to our infrastructure in our Dallas / Fort Worth, TX (DFW) data center to enhance stability and performance. The maintenance is scheduled to last two (2) hours, beginning on the morning of March 6 at 12:00 AM CST and ending at 02:00 AM CST. While connectivity during this window will be intermittent, some sites may be offline for as long as 90 minutes. As always, we will do everything we can to keep the interruptions to an absolute minimum.”
I called RackSpace Cloud to see if the email was legitimate. To RackSpace’s credit (again), the support center answered my call within 3 rings. And yes, the technician confirmed that the maintenance involved some storage array work. Separately, the RackSpace email to me mentioned:
“During this maintenance window we will be performing a series of software upgrades and disk space additions to our storage array to increase capacity, stability, and scalability. These upgrades will also allow us to further optimize our storage array’s performance, equating to faster data load times for your applications and better stability. During the maintenance period you may be unable to access, upload, or make changes to your website content. We will certainly do everything possible to limit the amount of time your account will be affected.”
RackSpace apologized for any potential inconvenience. And I must say: We’ve been pleased with the RackSpace Cloud. As part of a gradual migration, we moved each of our five media sites to RackSpace Cloud over the last five months.
Still, the cloud demands nonstop availability. The thought of a two-hour maintenance window — which could take some sites offline for 90 minutes — makes me shutter.
Read More About This Topic
Share This Post
Posted In: Service Level Agreements | Software as a Service and Hardware as a Service
Tags: Cloud Maintenance | Cloud Services | Rackspace Cloud | SaaS
Interact: Add a Comment | Trackback Link | Permalink
Subscribe: RSS Feed
Take a look at our 
Duh. Every IT operation, including Cloud, has some planned downtime for maintenance. God bless them for being upfront about it. This isn’t news.
Hank: I respectfully disagree with your assessment.
If Google went dark for 90 minutes, what would happen? How about Facebook? Twitter? Etrade? Ameritrade?
Thousands of businesses depend on Rackspace. I believe the potential for a 90-minute outage — during planned maintenance — is news.
-jp
Joe: while the theory of your assessment seems sound, you may need to better understand “your cloud” before assuming. These days “cloud” means any number of things, and it’s different to everyone. Are cloud generally designed to be redundant? sure, but to what degree? The statements you posted about Rackspace’s Cloud is specific the a single location, where it sounds like they are doing significant infrastructure changes. These changes may encompass all primary and secondary systems in a location, which could impact availability. If there is an assumption that there is geographic redundancy (take an entire location down and fail to another one) then I would say very few coulds do that today, though maybe the perception is that they all do. I’m not sure how long you’ve been a cloud customer, however it’s likely that from time to time any hosting provider will likely have to have a maintenance that will be service impacting.
So to your comment: “aren’t cloud platforms designed to be redundant, so that no customers go dark during system maintenance?” I’d say not neccessarily, they are designed to be redundant so that they can withstand failures.
Some of your comparisons are interesting. Google is a whole different animal, but if talking about only the cloud hosting piece, you might be surprised. Facebook and Twitter, I almost have to laugh; though they aren’t hosting cloud for customers, their are people that run businesses/apps with them. However, they don’t have a great track record for the last year in terms of going dark. Etrade/Ameritrade, well being financial businesses I would say take much of the uptime precautions into their own hands and spend hundreds of thousands of dollars (or more) to cover their redundancy.
This is idiotic – 90 minutes, really. Thats not even downtime. We are still searching for the Black Box that never requires anyone to do anything it just always works and does what you want it to do as fast as it can possibly be done without much thought from the user. A lot like this article.
Markets are not open on the weekends. I doubt eTrade or Ameritrade will be too terribly affected by the outage.
Coming from a Telco background, it still surprises me the tolerance level for downtime that IT service providers need. If this were an 800 number service, or other telephony service for that matter then any downtime would be unacceptable. There is always a technical solution for this problem – but at a cost. Are there cloud platform service providers that offer 5 9′s availbility? There must be a market for it.
Martin: I know there are a few service providers that offer five-nines guarantees. But planned service windows/maintenance aren’t covered by such guarantees… …
-jp
FYI, AutoTask often do this, usually on a Sunday. It’s annoying, especially when I had planned to review the previous week’s tickets. But it’s life.
Daniel: The day will come when such planned darkness isn’t part of life. Within 5 years?
-jp