Because the server room is certainly no place for pets
Machine hoarders burn cash – it's time to virtualise your legacy IT
Legacy IT is toxic. Virtualisation is the default choice for new data centre deployments, but for existing and legacy workloads, justifying hardware refreshes is often difficult. Shedding light on the often poorly-accounted-for costs hiding in your data centre can provide sufficient rationale to move your infrastructure forward.
People often romanticise legacy IT. Sys Admins fondly look at that old Compaq Netware server “pet” in the corner with a tape drive that hasn't been cleaned in a decade. Meanwhile, we make sure our indistinguishable-from-one-another modern data centre “cattle” are faithfully monitored, backed up and replaced.
It is assumed that because our pets have defied the odds this long, that they will continue to chug along, zombie-like, for years to come. Our “pets” avoid refreshes and migrations, and thusly spite the data centre gods.
As our other data centre tools have evolved beyond supporting these old platforms, and the people who built them retire, we mentally surround them with an SEP (Somebody Else's Problem) field. It’s time to own up to the real cost of running legacy IT (or rather, “Toxic IT”) in our data centres.
The real costs of legacy IT go beyond the one-time capital costs for deploying and migrating to new systems. The real devil is in the poorly understood operating costs and support debt that these systems incur.
Legacy systems require keeping staff or entire teams around "just in case" the dark arts of Novell or SCO UNIX need to be summoned. Support contracts maintained with application and ERP vendors who have long ago fired or let go the relevant developers are another challenge.
This opex is milking the cow of your operational budget to keep the system barely functional. You've reached the final form of these legacy systems when the support contract comes from a single developer.
Extended support on these old hardware platforms comes at significant cost, and the SLA's slip from four hour, to next business day, to "frantic eBay searches".
More obvious opex costs also hide in the power bills that those two-dozen legacy web servers rack up in comparison to their modern brethren. The cost of required software upgrades that are now licensed per socket (operating systems, databases, monitoring, back-ups), makes the “savings” of not buying hardware negative when you consider the licensing costs.
Consolidating 100 blades into 5 2RU hosts reduces cooling, power, real estate, back-up software, monitoring and management costs.
The least understood cost of legacy systems is risk. Everyone has their story about that poorly understood 10-year-old box dying, losing millions of dollars in revenue as operations ground to a halt.
These old systems require complicated (and often clunky) methods for replication, while modern virtual machines can be zipped to another data centre in a few clicks. Modern virtual back-up systems "just work" and allow rapid reliable restores in minutes.