I was asked recently to find a low cost setup for a small software development company (about 30 people) for their infrastructure that had growth potential. I came up with the following:

- One server running Open-E DSS software SAN/NAS: This holds 12 1TB disks in a raid 10 mirror and shares out the disks via iscsi, smb, and nfs (I am not using the other protocols available such as AFP, FTP...). It is certified for VMware. They use it for home directories and as an iscsi SAN for ESXi.

- Two servers running ESXi software attached via iscsi to the SAN.

along with a UPS, rack, switch, and firewall.

Last time....

In my last post I talked about some of the more common ways to get performance numbers out of your storage: hdparm, dd, and bonnie++.

These tools are pretty good at what they do: measure the performance of a single drive, single LUN, or single filesystem. For many sysadmins, that's all you need to care about. Now we'll look at some more advanced tools for measuring performance in different situations.

Why we need other tools

One project I've been working on recently is determining how much data we can push from a single host out to multiple shelves of disk storage.

Background

In my daily work, I've noticed that not a lot of attention is given to benchmarking various subsystems. When a new disk array comes in, there may be a run or two of bonnie++, but that's generally it. Network benchmarks? Generally I don't see people run them. Maybe you don't care; maybe there's an entire storage team that handles this at your site. However, if you do care, read on for my experiences.

Why benchmark at all?

The answer to question might seem obvious to some, but maybe not. I've often observed people at work getting requirements of X MB/s for a particular

I've recently had a number of projects at work that want to mix multiple volume managers on a single server, specifically ZFS and VxVM for SAN volumes (actually, three including SVM for internal boot disk mirroring). The projects generally are for database servers, and want to use VxVM for database volumes because ZFS currently has some serious limitation on database size (limited number of devices recommended in a single zpool) and performance (single threaded checksumming, for one). However, at the same time, they want to have access to some of ZFS's features (in particular, the ability to oversubscribe filesystems, dynamic resize, snapshots and rollback) for some of the other filesystems.

While taking part in a storage install the last few days, including creating and deleting quite a few RAID groups and so forth, I've had to click on quite a few "Are you SURE?!" dialog boxes, even for completely idle, never-used LUNs. Here's an idea for storage management folks: keep a bitmap of recently-accessed LUNs. Specifically, keep two; every five minutes zero the old one and flip them, then set a bit in the active bitmap for each LUN when it's accessed. Then pop up an extra "You've accessed this LUN in the last five minutes! Are you SUPER-SURE?!" scary box if someone tries to delete or in some other way imperil one of those.