Given all the attention Amazon gets when its US-East data centers fall over, it’s only fair to point out that Rackspace had its own stumbles last week, most revolving around its new “next-generation” cloud servers and its Dallas-Fort Worth (DFW) data center.
In a statement posted at 6:06 EST on Tuesday, Rackpace said:
At 2:10 pm CST, we observed a high number of dropped packets in an isolated region of our DFW data center. We found a portion of the virtual switches on several hypervisors were overwhelmed. We upgraded the virtual switches on the affected hypervisors which resolved the problem. During this time, some customers may have experienced network disruption.
Looking at all the incidents listed on the status page (below), it was a pretty rough week — the blue encircled “i’s” indicate an issue.
To be fair, Rackspace was up front about them. In a note to customers, Rackspace blogger Mark Interrante wrote:
“Part of being the open cloud company is being open and honest when we have problems. On Tuesday, we had a service disruption that caused network connectivity issues to our next generation cloud servers in our DFW region. While only impacting about 2% of our customers, for those that were impacted it was an unacceptable outage. We hate it when we cause these types of issues for our customers and will work tirelessly to earn back your trust and faith in us.”
The mystery of Amazon Web Services
As we all know, Amazon wants more businesses to put more workloads into its Amazon Web Services cloud. But enterprises like to know about the fiscal health of their major tech partners. The issue here is that Amazon makes it hard to assess just how much money its cloud business makes by burying AWS revenue in the “other” category on its earnings statements. “Other” includes “non-retail activities, such as AWS, miscellaneous marketing and promotional activities, other seller sites, and our co-branded credit card agreements.”
Below check out the revenue trend of “other” at Amazon since 2006 when AWS launched. And keep in mind that revenue does not equal profit.
Elsewhere in cloud — The Pivotal pivot, Heroku spiffs up marketplace
On Tuesday. EMC and VMware came clean (kind of) about the cloud computing spinoff that GigaOM first reported on in July. As reported, the spinoff– The Pivotal Initiative -- brings together GreenPlum data analytics, the Spring Java framework, Gem data caching and Pivotal agile development plus Cetas analytics services.
As it was described to me before the announcement — the spinoff would take on all of VMware’s “tier two” products leaving VMware itself to focus on Vsphere, but more importantly vCloud with which it is trying to extend its server-room virtualization dominance to the cloud writ large. And even huge vSphere proponents acknowledge privately that vCloud has issues there. The VMware statement announcing the Pivotal initiative was pretty basic: It didn’t specify what will happen with such miscellaneous stuff as Zimbra email, Socialcast etc. I guess we’ll hear more about that early next year.
Also, Heroku announced what it called a major update to its three-year-old Heroku add-on marketplace. As Venturebeat reported, the update offers better self-service capabilities that let developers using the Platform as a Service manage their add-ons more easily. And a new billing service will manage transactions and such irritants as declined cards and failed transactions, Heroku said.