SaaS Downtimes: Multi-Tenancy – Not the Cloud – is the Issue

Reading over the IT Project Failures blog over at ZDnet, the debate over the viability of cloud-based applications came up. Of course, it is the usual “universal outages” issue that gives people pause.

The main issue is that it has taken the biggest SaaS/Cloud companies –, Google etc. many years to get to where they are household names when it comes to on-demand technology. And since these infrastructures were built a decade ago, or longer, they HAD to be built under a multi-tenant architecture.

The more modern SaaS and cloud offerings are multi-instance and have far more separation between individual accounts. This diminishes outages in general, since the whole system sees less universal stress, and the dedication of resources like databases and application instances reduces the potential for universal outages of any kind.

I believe that soon, as the SaaS forerunners give way to second-generation, more adept cloud providers, the outage issue will become more and more a non-issue…

One thought on “SaaS Downtimes: Multi-Tenancy – Not the Cloud – is the Issue

  1. Martin,
    Thanks for that.
    Do you happen to know any “typical” figures for power and equipment consumption in a cloud vs. multi-tenant scenario? For example, say you can run 200 users for 5 clients/customers/companies on 1 instance of an application on a given server. If you convert to a virtualization scenario, you then have 5 OS stacks, and application instances, 5 database instances, etc. instead of one. BUT, you get the flexibility using the VM tools to scale each machine, and perhaps even move it to another server. So is it accurate to say that aggregate system resource usage could increase in going to a cloud scenario?


Comments are closed.