Wow..I hosted Tucows back in the early 90s at Phoenix.net....
Those days will never be forgotten Scott!
9 publicly visible posts • joined 24 Apr 2013
"NetApp has a solid base of customers that are happy with their products."
Don't forget that this is a 7mode base.
The Cmode shift, has only just begun with 8.3 raising the bridge over the moat. Give the base 3-5yrs to run into the EOL upgrade cycle and be forced to Cmode, to judge how happy they really were.
Gavin:
Is the idea that you cant fill ANY filesystem and not incur a penalty something previously unheard of?
It's no secret anywhere, it has not been for 50years of storage system evolution...and wont ever go away either.
The cost of mis-managing a storage solution has never been a hidden topic.
No SB..I think you're wrong, because this is a freshman year argument, it's specious and misleading.
You purchase an amount of storage. The 'usable' after raid/spares, etc, is a fair way to look at ROI. We all get that.
The PHYSICAL cost of an array based on what you physically get is under vendor control.
Past that, how you use it, they are no longer in control, and do not own the result. Green field destiny brother..you pick.
The is a point with all vendors, where you will incur operational penalties if you abuse the architecture. It's an immutable fact of having a filesystem..its a system.
To do what it does for you, it has rules, and there are no secrets in the storage industry at any level, what those are. Begin to fill it up, and you _really_ wont like it if you blindly expected the behaviour of it to remain static.
There is not a single filesystem that supports a transitory (data moving in, out, changes to data, etc) workload, that takes no penalty when you fill it up.
Sure, you could say "but what if I fill it, have no changes, take no snapshots, and only read from it" you can go farther with that on anyone's FS, but you still cant go 100% without accepting some form of penalty to your business at some point. At least short of a CD/DCV/Tape storage. But we're no talking about those.
The space available that you do NOT use, is your cost of doing business to maintain a steady, even, repeatable result to your users.
If 80% and no penalty is the line, and YOU decided not to lose 5% by going to 85, 20% at 90, and 75% at 98, and risk opening the door to an alternate dimension at 99.5%.
That is not the vendor's issue to price down. That is your unique dataset, with your unique workload, and that will, with your knowledge of what those limits are, deliver you your unique result.
This is your choice in how you administer your enterprise. And your vendor is there supporting you at ANY level you wish to fill it up to.
It isn't Hitachi's, EMC's, Netapp's, HP's...Microsoft's...nobodys.
Who here calls MS to complain that your PC slows down when the HD gets very near full, anybody?
No, you accept it as part of what filesystems require to perform the way they need to, to give you what you need. I -know- the moment I get too full on my desktop, I can feel it..and so does everyone else reading this.
The value that YOU place on resiliency and performance are your internal costs of doing business, and how you communicate WHY you simply cant use every bit and byte, is why you're employed as the Sr. Storage guy somewheres.
Can a vendor do better? Possibly. Should they..yes, indeed. Push them to. Tell them you expect it.
But to say 'it should cost less' because you cant use some of it, is a circular argument. Because at any price that the product is introduced to you at, you could say "It should be 20% less...because..."
But since you're already receiving a % discount off list with your vendors (who doesn't), you are already receiving relief from what you theoretically choose not to use to protect your enterprise.
And hey if you used almost all of it at that point, you got away with something for nothing.
With Regards..a faithful reader. :)