Could converged systems change the way that IT admins spend their time? Figures suggest that mundane tasks such as backups and restores and system patches take between two and ten hours a week for around a third of those responsible for administering systems.
Even more time is spent monitoring systems to ensure that they’re running smoothly. Automation would be a great way to solve some of these problems, but with administrators often having to juggle multiple, diverse systems from different vendors, that hasn’t always been an option.
Automation is certainly possible without a converged system, but it is more difficult. Bolting together compute, storage, and network components from different vendors into a system of your own involves the use of different interfaces.
Typically, a sysadmin will script tasks to carry out operations on each of these three components, but how many of them will have the time to write scripts that control compute, networking, and storage in unison, while monitoring and reacting to the feedback from all three? It’s doable, but for many, it will be a challenge.
The benefits of a converged infrastructure lie in simplicity for IT admins. Instead of trying to juggle different vendors’ systems while carrying out complex IT tasks, they can rely on a single, integrated system that manages a lot of things behind the scenes.
These integrated systems come in two broad flavours: converged systems, which feature components designed to work together and tested against a reference architecture, and hyper-converged systems, where the vendor puts a software layer over the top that handles all of the components invisibly to the user. In the first instance, the components can still be separated out and used independently. Not so much with the second.
What to automate
In either case, you’ll typically see a ‘single pane of glass’ management interface, designed to provide easy access to a variety of tasks that may have taken multiple manual steps in a different environment.
The marriage of compute, network and storage allows for the automation of many tasks that have typically been manual processes for IT administrators. One example is the patching of systems. This is something that may have required manual intervention before, because systems had to be put into a patch-ready state.
Software-defined approaches to the management of hyperconverged systems now make that easier to automate. Software can be scripted to cycle hosts into maintenance mode before applying the relevant system patches and spinning back up again.
The tight integration of computer, networking and storage in hyperconverged system also makes automated storage management pretty much mandatory. A virtualization layer needs control of the storage infrastructure to provide the kind of black-boxed virtual machine provisioning that hyperconverged systems offer.
These include dynamic provisioning and the automatic tiering of storage behind the scenes. Another task is configuring systems for load balancing, so that they handle varying applications with different workload characteristics.
Simply hiding mundane tasks from systems administrators isn’t the only goal here. Repeatability and extendibility are key. Manual tasks can be strung together into longer, scripted processes that will be handled the same way each time, and produce the same, predictable output. In a complex IT operation, the value of that consistency shouldn’t be underestimated.
Using this approach, IT admins can create repeatable workflows that handle specific task areas for the IT department, such as lifecycle maintenance. The latter means that a server is provisioned from a standard build, and that patches and upgrades are routinely applied. Replacing, archiving or deleting unused servers can also be part of that automated process.
System resilience and recovery time objectives keeping you up at night? Converged systems can be configured to automate backup and data recovery too. Snapshotting, replication and backup management can be scripted, taking a load off the sysadmin’s plate.
What does all this mean for business managers? After all, whenever an IT department can find a way to make an IT benefit explicit to the business, it should seize the opportunity.
The key phrase here is service delivery. IT organisations can exploit the automation capabilities of a converged system to offer more flexible, responsive services to business users.
One example here is the provisioning of services, rather than simply raw VMs. Creating VMs is a manual task. Creating a VM with associated applications and appropriate storage and network capacity is a service.
This too can be automatically fulfilled under converged infrastructure systems, because the tasks involved with provisioning - from bare metal bootstrapping through to allocation of logical resources - can be strung together and handled under one set of scripts.
This also makes it possible to create templates when provisioning resources for different workflows, so that users get the storage resource they need to meet their particular task set.
Developers, thanks to the use of APIs that can access IT processes, can access these processes programmatically. That makes them easy to build into self-service portals, or at the very least to define and access via systems management software that still leaves the IT administrator firmly in control.
Distributed systems management
Many of these automated tasks and processes are reactive – they help to streamline the tasks associated with incoming requests and problems. That’s great for IT departments trying to free themselves from a constant cycle of firefighting. But is that all it can do?
Some automation tasks within a converged or hyperconverged infrastructure are preventative. Consider system monitoring, which can detect emerging problems in the compute/network/storage ecosystem and then be configured to automatically create service tickets for them.
This automation could also be extended into other areas, such as capacity planning, so that sysadmins could be alerted to predicted capacity changes, for example.
This is perhaps the biggest advantage to automation, as executed in a converged system where all the components hang together cohesively: IT departments can use it to shift their focus. They can offload many of the mundane tasks that consume large amounts of their time, and then focus on innovating instead.
Setting up this automation will be a task in itself, though, and administrators must think strategically about what they want to achieve. And one of the biggest challenges here may be coverage. It’s unlikely for many that a converged system will be a greenfield implementation.
Many of these systems may be single appliances, designed for specific projects or locales. Broader converged infrastructures, incorporating multiple physical components from the same vendor, are similarly unlikely to be deployed in a vacuum. There’ll always been some legacy in there somewhere. Don’t get ready for the lights-out datacentre just yet. ®