Identify available storage monitoring tools, metrics and alarms
Skills and Abilities
Apply space utilization data to manage storage resources
Managing space utilization is easy with the data provided through vCenter, This data is located under storage view tab for any object, I prefer using it from the top cluster or datacenter level this will give the view for everything under that.
Under storage views there is 2 view types Reports and Maps.
Reports:
Will display the information in tables, depending on the information your showing it will show capacity, free space, snapshot space, space used, redundancy etc.
Reports View
Maps:
Displays the relationships between object like vms to datastore or hosts to datastore in a graphical presentation.
Spider web of a large system
Use these 2 views to get all the storage information you need to manage storage resources I could walk through what each column in each section means and how you can work with the data, But if you don’t know how you could use Free space data to manage storage requirements, then please close the vi client now.
Provision and manage storage resources according to Virtual Machine requirements
Storage is one of the most important resources as this is where the virtual machines live, The most common bottle neck I have come across when being called in to fix performance issues is Datastore latency, now this can come down to a few things but its generally because no one properly worked out what IOPs were required before assigning virtual machines to storage.
There is a couple of tools you can run and there is some estimate calculators out there http://www.wmarow.com/strcalc/ is a good tool to use or if you are P2Ving a Physical server you can run an IO meter over the machine to see what is needed. Rule of thumb is generally need between 50 – 150 IOPs per VM.
Before attaching LUNs to the ESXi cluster you need to conider 3 things
Space
IO
Reduntancy
Once you have answered those questions then you can configure the storage to meet the VM(s) needs
Understand interactions between virtual storage provisioning and physical storage provisioning
I’m guessing what its saying here is about Thin Provisioning and how it interacts with physical storage. Thin Provisioning gives the guest system the illusion of having more space than what is actually being committed Storage only becomes committed once data is written.
I may give a virtual machine 100GB drive and it has only written 5GB of data to that drive which means physically its only actually using 5GB of the storage, but has the potential to use the full 100GB. This then leads to over provisioning this is when you have assigned more storage than is available, I could have a 500GB Datastore but dish out 2TB of space. This is where alarms would come in handy as you do not then want all that space to be needed, that would end in tears.
Basically size the storage correctly, use the correct SATP and PSP for the attached storage, and your following best practice.
Configure Datastore Alarms
There is a handful of pre configured alarms that cover what most people will need.
These are view able from the alarms tab, they are only configurable from there they are applied so you will need to select the very top level vCenter to modify the default alarms.
These alarms can be modified or disabled and new ones can be created and configured and applied either to an object or a group of objects. so you could setup tier 1 datastores with different warning or alarm levels etc.
You can also setup what you want to happen when an alarm is triggered like snmp trap and or email.
Analyze Datastore Alarms and errors to determine space availability
Going on the above information this one is easy, if a datastore space usage is 80% it will show the warning, at 90% it will show the error, This will vary depending on how you have modified it, Selecting the datastore in questions and selecting the alarms TAB you can see the triggered alarms and will know from the alarms triggered what the space availability
Configure Datastore Clusters
1) Creating and configuring datastore clusters is pretty easy and straight forward. Go to the datastores view. right click on the datacenter level and select “New Datastore Cluster”
2) Will then be prompted with the new datastore wizard which will step you through the options. Here is where you can turn on SDRS.
3) Select fully automated or Manual for SDRS
4) Select SDRS run-time rules (not if SDRS was not enabled these option will not be available)
5) Select Data center and or cluster to apply this cluster to.
6) Now select the datastores you would like to be part of this cluster, Usually you would only use datastores that are viable by all hosts But here I have selected all datastores as I am just going to use the local datastores for this demonstration.
