There are several environments (at least in the beginning) that their initial infrastructure was not designed with virtualization in mind. When they started out - virtualization was something that you were asked to fit into the existing infrastructure - adapt your current processes and procedures, to this new emerging technology.
So most things worked, some didn't and we all learned during this time what the pitfalls were, how things have to change. But the essence here is - the infrastructure was not designed from the ground-up, with virtualization in mind. Some added a new storage array - to deal with the IO demand. Some upgraded to the 10Gb Networks to deal with the network load. Some added 3rd party tools to deal with backup / reporting / security or compliance.
Some datacenters are built from patches on top of patches on top of patches.
And eventually this will bite you in the butt, BIG TIME!!
This is why it essential to regroup - look at things from a different perspective and look with a different set of mind.- this provides the possibility to provide a better solution in the long term.
I gathered several members of our IT department and started a "think-tank". This group includes people from our storage, network, helpdesk and infrastructure group. The main purpose being, to invoke thought and discussion:
If we were to start fresh - how would we envision the infrastructure? If we could wish and get whatever we want - what would our environment look like?
I would like to share with you our journey (we are at the beginning) - because I think it would be beneficial for anyone to take a step back - and to look at the environment from a different view.
Several prerequisites for these discussions for the first stage were established:
- We start with a completely clean slate. No vendor lock-in, for anything.
- $$$$ - are irrelevant - it can be as cheap / expensive as we want.
- Once we have what the vision we would like to go with - then we adapt and fit it into our corporate policy's, politics, and budget.
This does not mean we will be able to implement all of our wishes, this of course has to be aligned with the business, its needs and its resources.
We started to to identify what components are parts of the infrastructure and below is the initial list we identified (I am sure this will grow during the discussions).
Let's go through each area and the questions we would like to address in each area.
- What do we use for the the core Infrastructure? Vendor? Technology?
- What kind of bandwidth are we going to be designing for?
- How do we incorporate BCP/DR into the network design? (Replication, VLAN design, etc.)
- Planning for IO convergence. Technologies - options - how to manage all this IO?
- What kind of disks will we be using? How do we incorporate Tiers into the storage?
- SAN/NAS? FC/FCOE/iSCSI/NFS?
- How do we incorporate BCP/DR into the Storage design? (Replication, LUNS, etc.)
- Backup/Restore. Which tools? What method?
- Which servers will we using? Rack mounts / Blades?
- Which Vendor/s ?
- Which Hypervisor fulfills our requirements?
- Is there place for more than one Hypervisor?
- Hypervisor provisioning?
- How is a VM installed? What are the criteria for new VM's? Importing? Staging?
- Reporting and Capacity - providing the proper insight to all levels in the company.
- How to maximize security for all layers?
- SLA - definition of what is supported, and how?
- How to present the infrastructure to the the end user without exposing too much? Provisioning, Chargeback/ Showback?
- Cooling / Racks / Management / Alerting / Power
- Providing redundancy on-site and off-site
- 3rd party facilities - including cloud (and its implications)
As I noted above - this is a partial list - and I am sure it will grow.
I actually look forward to working with my team on these discussions - it will be a good learning experience for us all and will help focus on what areas we should be focusing on in the near future.
If you have any comments on the components or on the process - please feel free to add them below.