Monday, 13 May 2013

Downtime for Storage #@#$!!++?!!

It is always challenging for an IT professional to take down production systems for maintenance. Storage systems compare to other IT systems, are the most critical parts of an IT Infrastructure and they always should be dealt sensibly. Planning downtime and maintenance window is very important aspect of any downtime. Always communicate all stakeholders (including end users) and take all the necessary approval from user department and IS function. Depending on the scope of the maintenance / upgrade designate a coordinator for all communication. Also take in account of services under effect, develop clear steps to restore these services and communicate the same accordingly among all stakeholders. Plan-B: This is my favorite part; I have always been great supporter for a backup plan. Sometimes such possibility is limited by inadequate planning. However you can always have a plan-b, it is better to have one instead of nothing. Start by performing a simple powercycle activity as per the system documentation, which may include all the important steps of disconnecting / killing active services and proper shutdown procedure. Once you restored the production services, start the actual activity. I hope this has been informative for you, please also share your experiences.

Thursday, 9 May 2013

SAN administration:

Before I got a chance to work on a NetApp box, I had many perceptions about the architecture and procedure difference with respect to other storage platforms. But when I started working on it. I found out that administering an Ontap SAN box can be very easy and fascinating task, if you have firm grasp of storage concepts because rest is a piece of cake. Storage provisioning based on customer requirement can be done easily by executing commands directly on Ontap CLI. You can execute multiple commands at once. It’s amazing to see that it took less than 2 hours for adding physical storage (DAE), configuring ownership, creating LUNs, assigning LUNs and iGroup configuration. Always keep all your commands in a notepad; this will help you for your next configuration. You can reuse these command lines; you just have to modify DAE and filer ids. I must say this is what they call rapid deployment. This is also true for any other process such as updating Ontap versions which take hardly 30 min as compare to other storage boxes which consume more than few hours for the OS update or patching.

Wednesday, 8 May 2013

Host Integration (Power Path memoir):

Let me introduce you with the lead actors. We have two HOSTs (Solaris) configured for a fresh DB RAC installation, a SAN box (VNX) and finally the protagonist EMC Power Path. Let’s configure raid groups using Unisphrere (the best interface I have ever worked with). Create some LUNs and put them in a new storage group. Connect both SPs to your two SAN switches, also connect you DB servers fibers to these switches. Now do the zoning, it is up to you what type you choose. I have chosen hard zoning for this configuration. Let’s install EMC host agent on both the RAC nodes. Verify the agent status on Unisphere, now the hosts can be added to the storage group. Scan hosts for the LUN visibility. Now install power path for multipathing functionality. Once you have installed PP. You have to make sure that the pseudo names and logical names of LUNs are same on both nodes. This is a mandatory requirement for RAC. However if both machines are freshly build and you haven’t fiddle with host integration, everything will go fine. But if you have assigned and remove LUNs on hosts, then you have to work hard to fix pseudo name matching issue. Yes there are several methods on internet for fixing this mess up, but none of them worked for me.