Ignore the pie-in-the-sky storage roadmaps. This is what's REALLY afoot

Roadmap! Roadmap! Bring out your roadmaps... I’ve now seen roadmaps going way off into the future and it is a pain. As soon as I start speculating about the future of storage, people seem to get very worried about breach of NDAs. But some general themes are beginning to appear. 1) Traditional RAID5 and RAID6 data protection …

[Buzzword]

Do you buy a new wardrobe when you have too many clothes?

Just get your users to clean up their data from time to time. Their own productivity will improve as they learn to chuck out useless info and only keep the important stuff; and you'll spend far less on storage costs.

This week the IT guys presented me with a (proposed) massive bill for new storage hardware, based on the fact that the current disks were 91% full. After a couple of hours of deleting redundant data, it was down to 35% full.

Just as we would throw away old clothes (or put them in a box in the attic), we should make more effort to throw away old data. If you keep buying new wardrobes for your end users, and not charging for them, then no wonder they just keep filling up.

[Fuzz]

Thing is your clothes stay pretty much the same size, data grows. Photos get bigger, more megapixels. Videos get bigger, HD higher framerates. Emails get bigger, used to be plain text now they're html with images in the signatures. More emails are sent if I want to keep a years worth of emails I need far more space than I did 5 years ago.

[Anonymous Coward]

>Emails get bigger, used to be plain text now they're html with images in the signatures

And that's one of the big problems with today's IT. I get emails from people with 1 MB of pointless images in their signatures. I want to know your name, your phone number, and what you do. That can be done in a few dozen bytes. All of these logos and certifications suggest to me that you've got no experience.

Once you get an email chain with the same signature ten times it starts to fill your inbox.

Most annoyingly no email programs seem to have a function to strip this crap out.

Now for a few comments on the article:

2) Software-defined is marketing bullshit. Yes, the clever bit's in the software, but then again it always was. ASICs are coded in software. The firmware that runs on them is software. Using generic hardware such as CPUs allows for quicker code rewrites. It always has. Anyone buying into software-defined being anything new is an idiot, or if I were being more generous, ignorant.

4) The biggest problem with automated tiering at the moment is it's reactive, not proactive. And granularity doesn't always tie in with data layouts. It's getting better all of the time though so everyone should be using it. Pick the best feature/price combination.

6) Not really. For now, Intel produces commodity CPUs which do the job. There's nothing stopping vendors using other architectures, such as ARM. x86 is just the best choice at the moment.

10) That's because people are frightened of Object Storage, particularly as there's no apparent common standard. Too much of a risk of lock-in.

[Just a geek]

Fully agree with point 7. I really can't see the point of FCoE. If you want to chuck data down ethernet cables then you can go iSCSI and 10GBE. If you want FC then you can go pure FC. Having a middle ground with FCoE just seems like a solution to a problem that no one has.

[Anonymous Coward]

Apart from iSCSI being Satan's own protocol. If Satan were in IT. Which he probably is.

[Anonymous Coward]

Pick your conspiracy theory:

1. FCOE exists because Cisco wants you to buy a boatload of switches.

2. FCOE exists because EMC is tired of customers reusing their existing FC switches and wants you to rip and replace the entire stack

3. FCOE exists because iSCSI is significantly lower cost and storage vendors would like to keep customers locked into thinking that storage is special and thus has to be expensive, when it's not and it doesn't have to be.

4. FCOE exists so that NetApp can finally do a decent block-access from their filers.

Did I miss any?

[Magellan]

Conspiracy Theories?

1. FCOE exists because Cisco wants you to buy a boatload of switches.

10Gb iSCSI does the same thing.

2. FCOE exists because EMC is tired of customers reusing their existing FC switches and wants you to rip and replace the entire stack

10Gb iSCSI does the same thing.

3. FCOE exists because iSCSI is significantly lower cost and storage vendors would like to keep customers locked into thinking that storage is special and thus has to be expensive, when it's not and it doesn't have to be.

10Gb iSCSI is not significantly lower cost than 10Gb FCoE. FCoE capable NICs are standard on most blade servers today, and there is not a significant difference in the cost of a 10Gb NIC and an a 10Gb CNA for rack servers. Most 10Gb switches available today are FCoE capable, and 10Gb ports on legacy switches are very expensive. That said, FCoE suffers from the same qualification matrix issues that FC suffers from, while iSCSI generally requires fewer compatibility layers.

4. FCOE exists so that NetApp can finally do a decent block-access from their filers.

Since FCoE is basically FC protocol over 10Gb Ethernet, if NetApp FCoE block access is "decent", than so must be NetApp FC block access. And NetApp iSCSI is widely used.