MirrorView /A Reserved LUN Pool

No you don't. I recommend you stick to RLP LUNs with a size of 10-30GB.

For example, let's say you have 5x2TB LUNs that you want to replicate. 20% of that is 2TB and that's your RLP size. So what I would do is reate 68x30GB LUNs (2040 GB) and assign those to the RLP.

Just make sure you don't exceed the number of RLP LUNs the Flare allows. See attached image for maximum number of RLP LUNs.

Hey Ernes thanks!

I have 41 free LUNs on a RLP on a secondary array, but I still have mirrors Admin Fracturing due to "Unable to create protective snap session on secondary array. Snapview returned an error on an attempt to create a protective snap session due to lack of free cache LUN's on secondary array. As a result of the error the mirror will be admin fractured. Add more cache LUN's on the secondary array and retry the sync request. (0x7152863e)".

I understand the criteria for RLP based on emc111716 which states:

- The Reserved LUN pool must contain at least one LUN for each source LUN. 

- EMC recommends a replication cache LUN with a minimum of 10-20% of the source LUN size for copy-on-first-write (COFW) data (including an overhead of approximately .2% of the source LUN size).

- In a failover situation additional LUNs must be available in the Reserved LUN Pool for the surviving SP; otherwise, updates fail to start.  Ensure that the secondary array has enough SnapView cache LUNs in the Reserved LUN Pool.

But, what about several 2 TB LUN?

Do I have to create (1) 400 GB LUN for the RLP in each array?

I'm confused with a numerous of differnet LUN sizes, I/O sizes and how to accomodate the RLP for them.

Fire away

I agree with Ernes, keep the LUNs fairly small unless you are running into the maximum RLP LUN quantity. If you have that issue, double the LUN Sizes you are using in the pool.

By the way, is the fracture occurring at a certain time of day/night? There are configurable alerts in Navisphere/Unisphere to notify of RLP running out of space. Do you have those alerts configured and are you seeing them?

It’s possible that one LUN has a large amount of change and consumes many of the RLP LUNs, essentially starving the other LUNs of RLP space.

What is the Array model and FLARE version? Newer versions of FLARE leverage RLPs a bit differently from older versions of FLARE (newer is better).

Richard J Anderson

First, you will want to find some time to upgrade the FLARE code on both of those arrays…

30.517 is the go-to code for the CX4 when FLARE30 is installed

26.026 is the go-to code for the CX3.

The RLP sizes, LUN quantities, and LUN sizes do not need to be the same on each array. You should size each array based on the expected RLP usage on that array. Take into account Mirrorview/A and local snapshots, as well as the maximum number of RLP LUNs supported by the specific array model. The CX4 can have more RLP LUNs than the CX3, so you may want to use larger LUNs on the CX3.

The primary concern for RLP sizing is the overall size of the data being replicated across all LUNs, and the change rate of those LUNs. If you use small RLP LUNs, you may see that a large/busy LUN uses many RLP LUNs and smaller/less busy LUNs may only consume one RLP LUN. This is normal behavior. You don’t ever want to run out of RLP LUNs. Also, each LUN being protected owns its RLP LUNs, no data LUN can share a single RLP LUN with another data LUN, that’s why you need a fair number of RLP LUNs.

Good news is that multiple snapshots and replications on the same data LUN share the same RLP LUNs.

Richard J Anderson

Thank you Ernes and Richard,

Fracture occurred late last night around 23:00, but other times (twice before) at random times.

I have an alert informing me when a fracture has occurred but not when the space is limited or near limit.

The primary array is a CX4-480 running 4.30.000.5.509, the secondary array is a CX3-20c running 3.26.020.5.020

Now could you please help me answer:

- Do I have to have the same amount and sizes of LUNs in both arrays?

- Should I take in consideration individual LUNs to create my RLP or the overall (added) size of LUNs?

- Is it normal that Replicas ---> Reserved LUN Pool ---> Allocated LUNs tab data rows are empty on a CX3 array in Unisphere? Because Replicas ---> Reserved LUN Pool ---> Free LUNs tab displays free LUN data.

I appreciate your help!

Manuel Rios Tamez.

Ernes, In regards question 3, what I meant was that information about Allocated LUNs in a CX3 array is unavailable; Unisphere just displays the page with empty lines.

If I click on Free LUNs tab, a list of Free LUNs is displayed.

I have no issues with a CX4 array on Allocated or Free LUNs tabs.

Ernes and Richard,

Let me figure it out a game plan for my environment.

I understand every environment is unique, but at the same time there are some settings that are similar or kind of standard; if I reply in a couple of days with some preliminary LUNs and RLP sizes, could you please take a look and let me know your opinion based on the numbers?

Best regards,

Manuel.

Sure. Just post the question here and I'll give you my 2-cents as soon as I can.

I'm pretty sure you need a minimum size of one RLP that is .2% of the source LUN size - this was based on an older version of Snap - it may still apply. Try creating one Pool LUN of 400GB and see what happens.

glen

I'm working on some numbers for my environment.

I really hope you guys can point me to the right direction, make sure they at least make some sense.

Manu.

Since you are replicating 29TB you have no choice but increase the RLP LUN sizes. You can have up to 128 LUNs in the RLP I would recommend you go with 95x60GB LUNs in the RLP. You can also decide to go with 113x50GB LUNs but that will only allow you to add 15 more LUNs to the RLP since you will reach the limit. By going with 60GB LUNs in the RLP you will allow yourself room to grow since you can add 33 more LUNs to the RLP.

One thing to note is that once you create a replication session for the 1GB LUN that 1GB LUN will use a 60GB RLP LUN. You are not allowed to choose which RLP LUN is assigned to which source LUN. There is a way around that but it's very tedious work. For example, let's say you want to replicate 5x1GB LUNs and 5x1TB LUNs. Create a RLP with 5x1GB LUNs. At this point your RLP only contains those 5 LUNs. Now configure replication for the 5x1GB source LUNs only. Verify and make sure all LUNs in the RLP are being used by the mirrorview sessions for the 5x1GB source LUNs. Now configure the remaining RLP LUNs and configure the other replication sessions. If you don't do it this way then the 1GB source LUNs will take a 60GB RLP LUN which is a waste. It's a decision you have to make.

Hope this helps. I'll also be interested to read other replies as well and see what others think about this configuration.

Out of curiousity, what's on the 1GB LUNs? Are those quorum drives? Do you really need to replicate quorum drives?

Hello guys, please take a look to the below info, and let me know if those numbers make sense or there is something I should change.

I really appreciate your help and your input!

Manu.

-------------------------------------------------------------------------------------------------------------------------------------------------

-------------------------------------------------------------------------------------------------------------------------------------------------

Does the total LUNs in a array includes Private LUNs and obviously RLP?

Am I limited to my smaller array, in my case CX3-20c with 128 total LUNs for mirror LUNs, and RLP?

Private and RLP (metaLUN component LUNs) are all counted as LUNs.

glen

Does the total LUN count in the array includes Private LUNs and RLP?