What’s wrong is that “Best practices” is a misnomer. What’s perfect for a commercial data center can be totally insufficient for a cloud service provider. But there are steps we’ve seen repeatedly and can all learn from.
The first step I call “Basic Practices.” When organizations install a FlexPod, a Vblock, or some other type of converged infrastructure solution, they start using all the administrative tools that come along to manage all the pieces. And wow, these can be exciting. Using service profiles in UCS Manager to move BIOS settings, MAC addresses, and network security along with an operating system from one blade to another in one step? That used to take days, and now we can switch a blade from function to function on a schedule!
But those basic practices start falling down as the newness wears off and the experts who built the system go on to their next project. And system configuration changes. And something goes wrong. Now we’re either “getting the band back together” (but Jake and Elwood are on a state project for the next 3-5 years and aren’t available) or we’ve got the one man band playing all the instruments. Picture that poor guy trying to poke around in UCS Manager and vCenter and SSH-ing into a load balancer context and trying to focus on solving the problem while re-learning all the interfaces.
Zenoss has been helping people move beyond that problem for years by unifying monitoring for all the equipment in the data center into one system. And that’s our “Best Practice” recommendation for a converged infrastructure deployment. Make sure that your gathering availability, performance, and event data for the virtualization, compute, storage, and network components of the converged infrastructure into one system. Then your one man band can efficiently navigate from technology to technology. And if it turns out you do need to bring in a larger band, they’ll all be working from the same arrangement.
Unifying operational data in one system has worked time after time for people across multiple industries. In fact, if you’re not doing it you’re operating at a competitive disadvantage with a higher cost structure or worse availability.
But what about “Leading Practices?” You can (and should) go beyond “best practices” to adapt your IT to your organization’s business needs. Here’s a few that our customers have chosen:
- Tenant service orchestration and tenant-focused consoles
- Provisioning-driven monitoring
- DevOps Automation
- Service-based infrastructure management
- And finally, Project-focused monitoring vs Centralized monitoring – hmm, looks like I need to write about this one!
What’s the summary? Simple. If you don’t have a single software tool that manages every component in your converged infrastructure you’re not responsive enough to problems and you’re spending too much. Get there quickly and then figure out which Leading Practices make financial and operational sense.
Other Posts in this Series
- An Expert’s Take on Converged Infrastructure Learning Curve – Part 1: What the Business Wants
- An Expert’s Take on Converged Infrastructure Learning Curve – Part 2: The Trouble with “Agility”
- An Expert’s Take on Converged Infrastructure Learning Curve – Part 3: Two Very Naughty Words: “Chargeback” & “Showback”
- An Expert’s Take on Converged Infrastructure Learning Curve – Part 4: How to Talk About “The Squeeze” of Converged Infrastructure