Prepare for a growth spurt when you virtualise systems
Adjusting the scale
Virtualising systems often means scaling them up. Sometimes, disparate networks of machines are consolidated together, creating a mega-portfolio of assets. This carries a special set of technical challenges, but let’s not forget the managerial ones.
What happens when you scale a system by virtualising servers and cramming more of them onto more powerful hardware?
Your IT team finds itself having to manage a bigger volume of virtual machines, with all of the associated networking, performance and maintenance issues that it entails.
“If you are scaling and making things that you have to manage bigger, you will be stressing the processes and the individuals that do the work,” says Phil Everson, technology partner at Deloitte.
“It’s not just the case that you will need more analysts to cope. It may be that you can’t manage it in the same way any more.”
What the head of storage architecture was able to do in a relatively small-scale system may change when that architecture suddenly covers several networks in different geographies.
If you have 50 people handling storage management across three data centres on two continents, it is not just the technological challenges that become more complex.
“Demand for scaling up from the business requires a much more mature bilateral set of communications,” Everson says.
“People have been trying and getting a bit better at it over the past decade, but it’s still an enormous challenge.”
This focus on solid, clear agreements with the business takes more rigour than you might think.
In the frame
“Does an agreement mean that it has been approved, or accepted? Does that mean there’s budget to do it? That there are people actively working on it and committed delivery dates?” Everson asks.
“Often, it doesn’t, and both sides miscommunicate as a result.”
Simple techniques, such as producing a register of all the projects currently being delivered in IT, can help here. Such a register could include a description of all the milestones planned for the next 100 days.
Management frameworks such as ITIL or Capability Model Maturity Integration can help organisations to manage infrastructures by refining processess. However, they are not a substitute for good IT management.
“They are an approach that drives the right nomenclature to enable good management. They don’t ensure that you have it, but they can help,” Everson says.
Automation of manual processes can be a huge help, reducing error and cost
The same goes for automation, according to Brian Murray, principal consultant at IT services company 2E2. Manual processes are costly to implement, and as a system scales up staff become overworked and the cost increases.
Automation of manual processes can be a huge help, reducing error and cost. Processes ripe for automation include capacity management and automated storage tiering.
However, automation can backfire if these processes are detrimental to your IT management environment in the first place.
“If you are automating a process that is clunky and slow, then all you end up with is more of these problems, more quickly,” Murray says.
He highlights one process that can often be malformed from the start: patching and provisioning. This can often be done without proper patch testing, or it can be so bogged down in bureaucracy that it takes far too long to complete.
“If you look at how people make change management processes, mistakes are made, things are out of date, and it reduces confidence in that process,” Murray warns.
Getting some of the basic tasks right can give you a solid foundation on which to automate other processes. Configuration management can be mired in bad process, but if you get it right, and automate it effectively, it opens the door to a lot of benefits.
Automatically updating your configuration management database paves the way for more effective infrastructure management.
The management problem is likely to become less tractable as we move further towards public cloud-based services, many of which may be mixed and matched from various vendors.
“When people moved to more modular managed services from large outsourcing deals, they had more contracts to manage,” says Everson. “That led to more finger-pointing when things went wrong.”
He believes we will face the same problems when we move to cloud-based services unless we have a solid management framework in place.
Overall, a focus on proper management at the outset as companies scale up their computing infrastructures will save a lot of headaches later.
Complex frameworks and rulebooks will get you so far, but you still need to practice management techniques on the ground, ideally including a mature relationship with the user base. ®
Sponsored: Hyper-scale data management