If you ever suffered from unplanned or even planned downtime because of storage failure/maintenance/migration, you probably want to prevent risk of any downtime. You can achieve it by using Vplex Local and even more with Vplex Metro.

 Vplex – what is it?

Vplex is an array acting as a bridge between block storages at the back end and servers on the front end.

Vplex creates virtual volumes from the Luns provided by the storage and presenting them to the servers. Using Vplex cache improves your performance and mirrors your data to local and remote storages.

You can connect any storage from any manufacturer as long that it has fiber channel connectivity.

Vplex is a high end array, using 1-4 engines, each of them with 2 active/active directories, meaning no single point of failure.

You can use Vplex as Local – by connecting one or more storages to one Vplex array or as Metro by connecting one or more storages to each of two Vplex arrays connected to each other.

Vplex Local

I have storage with high availability already, what do I need Vplex for?

By connecting one storage to Vplex you achieve 2 things:

• Better performance because of the Vplex cache.
• Migration of data between storages – for example when replacing storages – with no downtime.

By connecting more than one storage to Vplex you achieve even more:

• You can create active\active Lun mirrors between 2 storages meaning that even if one of the storages is down for any reason, data is still available for servers with no downtime.
• By using Vplex mobility, you can move data from one storage to another non-disruptively in order to use the most optimal resource utilization.
• Vplex uses high end hardware build on Vmax architecture with no single point of failure.

Vplex Metro

Vplex Metro is the next step of high availability, moving from five to seven nines and even more.

By connecting two Vplex arrays, with a storage connecting to each one of them, you can create mirrors between two remote system, ensuring that data will always be available to servers.

The way Vplex Metro achieves that is by making the cache between the two Vplex arrays synchronized, allowing active/active read & write on both of them.

In this case, storage failure or even Vplex failure will not cause data loss or loss of data availability.

As expected all Vplex Local capabilities are preserved on Vplex Metro configuration.

Using Vplex Metro on 2 Sites within a campus.

The best way to reach the maximum out of Vplex Metro is by using it on 2 sites within a campus.

Actually what it means is this: you can stretch the site capabilities into two different sites having FC connection between the two sites.

What does it mean?

When using Vplex Metro between sites you have the same Luns on both sites active/active for read & write so from clusters view you are on the same site.

Some key features to achieve the maximum from using Vplex Metro on two sites.

• Create ISL (Inter Switch Link) between sites on FC switches (use 2 different domains on each site).
• Separate compute resources between sites, meaning for each cluster (Oracle, Microsoft, VMware, Hyper-V etc.) have at least one server on each site.
• Connect (zone) all servers to both Vplex arrays and expose virtual volumes from both. You don’t have to worry, the multipath software will know how to deal with it. (Power path is preferred but any multipath software will do).
• Deploy Vplex Witness on third site to avoid data corruption in case of disconnection between sites.

Possible scenarios with Vplex Metro on 2 sites.

1. Storage planned or unplanned failure – all servers will continue to work from the 2^nd Vplex array. Once the storage is back up, the data will be automatically synced and back to normal.
2. Vplex planned or unplanned failure – all servers will continue to work from the 2^nd Vplex array. Once the Vplex is back up the data will automatically be synced and back to normal.
3. Planned site maintenance – move all virtual servers to 2^nd site and continue to work from that site with all servers till maintenance ends. Once site is back up the data will be automatically synced and back to normal, virtual servers can move to original site once it’s up.
4. Unplanned site downtime – all virtual servers from the failed site will start on the 2^nd site according to cluster high availability configuration. All physical servers on 2^nd site will continue to work. Once site is back up the data will be automatically synced and back to normal. Virtual servers can move to original site once it’s up.
5. Disconnection between sites:

1. Both sites can reach the Vplex Witness. According to preconfigured configuration, the Witness will configure one Vplex as read/write and the other Vplex as read only. Once connection is back up, data will automatically be synced and both Vplex will get read/write capabilities.
   1. Only one site can reach the Vplex Witness – in this case the Vplex that can reach the witness will become read/write and the Vplex that cannot reach the witness will become read only. Once connection is back up data will automatically be synced and both Vplex will get read/write capabilities.
   1. Both Vplex are connected but can’t reach Vplex Witness – in this case an alert will be created but all Vplex Metro capabilities will continue.

Customer story

One of my fun installations was with a customer who had 2 sites within his campus. The customer used Vplex Local connecting 2 storages, each of them in other site.

After discussing a technical refresh – the storages were quite old – we planned to grow to Vplex Metro with 2 new storages. The plan involved breaking some walls in one of the sites, in order to get more space for the new equipment.

So, we did all the new implementation on the 2^nd site (where the Vplex Local was already installed) then a shutdown was done for the 1^st site and the customer continued to work from the 2^nd site with no downtime. Walls were then rebuilt, electricity was arranged, and the site was ready to receive the equipment.

One storage and one Vplex were shut down and got transferred to the first site. After relocation, the storage and Vplex got connected and brought back up, data was automatically synced, and the customer was again working from both sites.

The fun thing in all this story is that no downtime was experienced throughout the whole process.

Replicating Vplex Metro to DR site.

In order to achieve the best replication capabilities to DR site with Vplex Metro look for my next blog talking about “DellEmc Recover Point” with Metro Point designed especially for Vplex Metro.