Storagebod Rotating Header Image

Excessive Sorrow Laughs….

So you’ve manage to get yourself a job in storage? Commiserations, why did you do something so crazy; I hope you enjoy pain and misery, this is now your world. Perhaps if you do a great job, you’ll come back as something higher up the food chain such as desktop support.

Anyway here’s some hints and tips

1) You will never have the right amount and type of storage but it is probably better to have too much than too little. Applications fall-over when they run out of storage and adding new storage into an environment at a run is never a great deal of fun. Don’t believe vendors when they tell you that it is non-disruptive; even if it is non-disruptive technically, it will always be disruptive in other ways that you do not expect.

Learn to be economical with the truth about what you actually have available; keep something in your back-pocket for a rainy day but don’t save the day too often. Miracles need to be the exception as opposed to the rule.

2)Related to this is the fact that no end-user has any actual idea of how much storage they will use. They will glaze over when you start talking about terabytes, gigabytes, exabytes; my experience recently is that they under-estimate but this is probably a factor of the sector I’m in.

3)Every best practice document appears to have been written by someone who has shares in a storage company. This is especially true for databases; you have various options…

  • smile and allocate what they ask for
  • smile and tell them that you’ve allocated what they’ve asked for
  • frown and have an argument
I’ve been around for long enough to know that the last option maybe the most tempting but it only leads to pain.

4)Non-disruptive upgrades are rarely so; read the small print as to what non-disruptive means. Code upgrades will always result in more work for every other team as opposed to the Storage team as they struggle to bring their environments up to scratch to meet the crazed requirements of your chosen storage vendors.

5)Fibre-channel is not a standard; it is a vague suggestion of how things should work. Hence point 4)! But Fibre-channel scares the crap out of people, you start waffling on about FLOGIs and you can get away with murder. (Serious hint, don’t mix up WWPNs and WWNNs…understand the difference..please!)

6)Of course you will be tempted to head down the NAS route; whatever you do, don’t mix NFS and SMB shares…every vendor claims that they have a solution to the inherent problems with the mixed security model. They don’t! It breaks in subtle ways and never underestimate the power of a Mac user to put some very strange things in a filename.

7)’But I can buy a 1Tb USB disk in PC World for £50′; learn to tune this statement out or you will be committed or jailed.

8)Everyone can do your job better than you can….until it goes wrong. In your darkest hours, remember point 4); there is nothing more joyful than realising that a single storage upgrade can mean many hours of disrupted lives for every other team.

9)There is always a better technology; you just didn’t buy it. Don’t worry about it; what you’ve got will do most of what you want and probably most of the time. This is why the same sales-guy you bought NetApp from will later turn up selling you EMC; they aren’t clever enough to understand subtle differences in technologies…so basically, they are selling you a different badge.

10)Storage is actually quite easy…everything which surrounds it is hard…

11)Learn to laugh….


2 Comments

  1. K.T. Stevenson says:

    Too much truth! Thanks for the laugh.

  2. Ed says:

    Counter argument to 7 is the unrecoverable bit error rate. You can probably find the “raid 5 dies in 2009” rant easily enough. But the two orders of magnitude in reliability is a big different. I also find 2/3 applies to iops and megabits as well.

Leave a Reply to Ed Cancel reply

Your email address will not be published. Required fields are marked *