Maintenance Sucks

Over the past week, I have been involved in multiple debates regarding the role of “maintenance” in establishing an effective software business. In one conversation with a key analyst at a top tier research firm, I listened to the analyst proclaim:

“Given the last 15 years of end user maintenance experience, customers have concluded that maintenance sucks. Avoid it if you can because something always breaks.”

Contrast that sentiment with Aneel Bhusri's (President of Workday) comments on a panel hosted by David Dobrin of B2banalysts:

“As a software vendor, the SaaS model is critical because it enables Workday to deliver updates and upgrades in a cost effective manner.”

If “maintenance sucks” and it should be avoided, why is Aneel so keen on efficient maintenance as a central tenant of the Workday business model? I believe I can bridge the gap between these statements by adding a couple of clarifying clauses:

“Given the last 15 years of end user maintenance experience [with legacy multi-platform server applications running on a bloated general purpose OS], customers have concluded that maintenance sucks [because, despite claims of certification, the patches from each vendor are never really tested in the exact configuration that we have deployed]. Avoid it [maintenance] if you can because something always breaks. [Unfortunately, customers that avoid maintenance create a miserable situation for a software vendor because the value of the application becomes marginalized while the vendor cost of maintaining legacy code on multiple platforms escalates]. As a vendor, the SaaS model [or a virtual appliance model] is critical because it enables Workday [or any other application vendor] to deliver updates and upgrades in a cost effective manner.”

When you pull these 2 sentiments together with a bit of editorial glue, they stick. SaaS is booming because it enables the vendor to deliver the application in a more efficient manner, and the customer gets application value without IT hassles. Salesforce.com is able to deliver their application to customers with just 8% of revenue for R&D compared with the more typical 16% that is standard for the application industry. I can only assume that Workday will achieve similar results due to the efficiency of maintaining a single code base running on a single platform for all customers.

Virtual appliances, done correctly, yield the same benefit. The vendor assumes responsibility for all software maintenance. The customer accepts maintenance (and presumably value) without hesitation because it is tested by the vendor before it is released to the customer – in the exact configuration used by the customer. The vendor and the customer get the added benefit (compared to SaaS) of flexibility regarding the network upon which the application is hosted. A virtual appliance is portable to any network that exposes a hypervisor host. The vendor can host it. The customer can host it. Or it can run on a service like Amazon's EC2 or IBM's pending Blue Cloud.

The concept must be gaining traction because the IT alarm sirens are shrieking:

WARNING! WARNING! VIRTUAL APPLIANCES ARE COMING! THEY MIGHT BE ON YOUR NETWORK ALREADY! THEY ARE BAD BECAUSE YOU LOSE CONTROL OVER THE TECHNOLOGY! THERE ARE SECURITY ISSUES! THERE ARE MANAGEMENT ISSUES! STOP THEM BEFORE USERS EXPERIENCE HOW EASY THEY ARE TO USE!

These are the same warnings that sounded regarding SaaS about 3 years ago. Anyone want to bet that the outcome will be different? My advice to the IT status quo – don't fight virtual appliances. Just like SaaS, customer value will trump IT control. Work with the emerging vendors to get what you need while giving your customers what they want -- application value without IT hassles. Maintenance doesn't have to suck.