Posts by swissarmyknife 11 publicly visible posts • joined 10 Aug 2015
Agreed, it's not an IF they will fail, it's a when. And if more people understood how miraculous it is that they even work to begin with, the thousands of possible failure points, there would be a LOT more emphasis on BC/DR, and a lot less of these events. I eventually got to where I couldn't sleep at night anymore thinking about all the possible ways things to go south, and the older they got, the worse the stress, because you know the lifespan of a LOT of the parts is 5 years, and they're installed in large batches. If you have 100 drives installed, and they were from the same lot, and 3 fail at month 50, guess what is going to happen next month? And yes, a lot of shops hold on to that hardware longer than the recommended 3-5 years.
Blades still have a lot of space to work in. Convergence is great, Hyperconvergence even better. But only a few vendors are doing the size as needed on storage portion of it. In that area, if you already have a sunk investment, blades are still VERY efficient. Convergence works where the guests are small and medium sized. It becomes less financially attractive when the guests are large or huge. In those cases, blade centers, attached to SAN is still a great bargain. And when you pair it with systems that can last 10 years (a.k.a, IBM HS), you have a long term investment that has very low operational costs per year, and very seldom a large capital outlay (especially if you use storage that is not forklifted, but older components replaced with newer components without tossing the whole baby out).
As moving the goal...which is what this does. Is there a business case for doing this? Yes, several. Is a clickbait title of hybrid is dead, or some other such stupidity immediately going to make me no take you seriously? Yes.
Hybrid arrays, flash arrays, yes they're expensive. But you're moving the problem. If I have a hundred hosts, then you can easily do the math and see where this is a bad decision. I'm using expensive Ram (not horribly expensive, but expensive enough), then I'm adding software licensing. I'm buying once for that array, I'm incurring the pernix cost, every SINGLE time I enable a host to do caching.
A few hosts? Sure, that's not a bad idea. Enterprise-wide? Don't be daft. You're looking 8-10k per host, all costs combined. By the time I hit that dozen mark, I'm thinking that 50k add-on to my array is looking a lot better. By the time I hit 20 hosts, that all flash array looks charming.
Kevin and Trevor are merely pointing out that if you buy the vendors numbers, or even gartners, you're buying a scam. I've used gear that got great synthetics, but was an actual dog on a real workload. And the opposite. One of the cheapest arrays I've managed, certainly one of the smallest (A storwize v7k), is shockingly fast...but got mediocre reviews and numbers from the big reviewers. But on a real virtualized workload (random as hell), it's phenomenal. We all have our pets, and companies we hate, but ALL of them slant the numbers and modify the systems to make the best of the synthetic tests they KNOW will be used on it.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
