Quoting Jim Gray [dead link] again:

“We have an embarrassment of riches in that we’re able to store more than we can access. Capacities continue to double each year, while access times are improving at 10 percent per year. So, we have a vastly larger storage pool, with a relatively narrow pipeline into it.

We’re not really geared for this. Having lots of RAM helps. We can cache a lot in main memory and reduce secondary storage access. But the fundamental problem is that we are building a larger reservoir with more or less the same diameter pipe coming out of the reservoir. We have a much harder time accessing things inside the reservoir.

You really ought to read the whole interview [link dead]. It’s very interesting.

Of course, I don’t really have a clue on that subject but as the discussion moves to access times (currently at 5 ms which is still very long!), I always wonder why those HD manufacturers don’t put multiple heads on the arm… For example, if you’d put 4 heads on each arm, you’d have to move the arms 4 times less to access any cylinder on the disks. Considering disk rotation to the right block marginal (remember at 15 000 rpms a half rotation only takes 33 µs!), that would almost divide access times by 4 instantly!

I cannot even imagine this being a cost issue… so, what am I missing?


Comments from long ago:

Comment from: curio

Un autre article assez intéressant bien que datant de 2000 nommé “Le disque dur le plus rapide du monde”

2003-07-30 20-05

Comment from: chro

Your brain too, can access conciously at one moment only a very small part of it’s memory. And latence for getting back some information is sometime measured in hours or days.

slow pipes are not such a big problem, at least with the usual usage of our mass storages systems.

expect progress in the way we index datas. (WinFS?)

2004-06-03 19-36