Beside my previous post regarding VMware HA Agent – Lack and Limitation and the recent series of articles about Hardware Health Monitoring via CIM written by Charu Chaubal himself , I wanted here to give an overview of what could be done, in my humble opinion, to enhance some of the VMware vSphere’s core features.
To summarize Charu Chaubal’s articles, CIM, which is a subset of the WBEM tools, allows multiple parties to exchange management information about these managed elements and also provides means to actively control and manage these elements. What does that mean in plain words? If you buy different servers from four different vendors, and networked them together, using CIM you can view the same information about each of the devices, such as manufacturer and serial number, the device’s model number, its location on the network, its storage capacity, and its relationship to the applications that run throughout the network.
There is another standard for managing hardware called IPMI which Charu will address in future posts I believe. IPMI is a controller on chipset (BMC) that monitor system health and manage the system locally and remotely. Although those two technologies, WBEM and IPMI, have overlapping specifications, the goal is to provide better manageability of your infrastructure. For the sake of this article I call this the Infrastructure Information.
So what’s the point with VMware DRS/HA/FT? VMware DRS, HA and FT, are good technologies! But if they were using WBEM/IPMI they would be great technologies.
Imagine that based on the Infrastructure Information retrieved from WBEM/IPMI, VMware DRS/HA/FT could intelligently place and load-balance VMs across your Cloud (VMware Cluster), whether it’s a private, public or hybrid Cloud.
What kind of Infrastructure Information I’m talking about here? Any information that is required to create snapshots of your hardware health, status and performance. Something like sanity checks that VMware DRS/HA/FT will use to star the hardware.
- A fully redundant server, that is with dual- PSU, NIC, HBA, FAN, is rated with more stars than a server without such redundant hardware.
- At instant X, a healthy server would get more stars than a server suffering high ECC or TX/RX errors.
- Newer server with high-end Intel processors would be starred compared to a 5 years old server although it was cutting edge hardware at that time.
Then we can add another level to unleash the power of VMware DRS/HA/FT. Imagine that you could star your VMs per level of SLA that you define in vCenter Server. That would be an extend to the current options available to VMware DRS/HA/FT. Again for the sake of this article I call it Operational Information.
- Bronze equals RTO:96/RPO:48/no HA/no FT/2 stars rated hardware min.
- Silver equals RTO:48/RPO:24/HA/no FT/4 stars rated hardware min.
- Platinum equals RTO:4/RPO:4/HA/FT/5 stars rated hardware min.
With the dynamic Infrastructure information collected through WBEM/IPMI and the Operational Information Policies you put in place, you could bring VMware DRS/HA/FT great technologies to the next level, that is vCloud! I don’t know VMware’s plans for future releases of vSphere/vCenter, but I’ve been told that a lot has been done to enhance those features, HA for instance, as per Duncan’s reply to an earlier post.