Its autumn. The nights are drawing in, the mornings are colder, the leaves are turning brown, the squirrels are preparing for the winter; burying acorns in the lawn, building a store of tasty bites for the spring.
Data storage managers are like squirrels too. Our acorns are tasty bytes of information and we store them with serious intent. Without them most human endeavours would stop. Just like that.
Data storage is like electricity, or water or oil. It is a staple of modern life. And creating it, making it, selling it, supplying it, implementing it, and managing it have all become more difficult.
Wed like it to be delivered like water from a tap or electricity from a socket but thats not our world. Our world is behind the socket, upstream from the tap. Were the equivalent of electricity generating stations, distribution networks, transformers, 3-phase power supplies. And our world is getting vastly more complicated.
Yes, we like to have the technology, but why does it have to be so complex?
It used to be so easy
It used to be a fairly simple business. Computers had hard drives which stored files as directly-accessible blocks of information. We backed the stuff up to tape and buried these in vaults, just like the squirrels bury acorns in the grass.
Then the first tectonic data shift occurred as the single data storage continent broke up into three:
- Blocks served over a Fibre Channel network SANs,
- Files served over a network NAS,
- And the original direct-attached storage - DAS.
But now, this tripartite split is shattering into dozens of sub-specialisms. We dont have the odd storage technology island any more, we have whole archipelagos.
Its inevitable. Storage has become a collection of specialisms, a glorious array of acronyms. How can any one person understand it all?
Wrong question. Why should any one person want to understand it all?
We dont expect a brain surgeon to understand heart transplant techniques. Thats where storage is going; towards a collection of specialists each working on part of the storage problem and managed by a storage director or similar function.
Why have specialists? Because specialists do things better. You get better results for less money in a shorter time. Its the way we humans work: we make things better, we find fault, we improve, we develop.
Whats a poor storage person to do?
Embrace the complexity. Welcome it. Its a good thing. It means we get at more data for less money, we store it for less money. We protect it better for less money. We deliver a better storage service to our users.
Interfaces are getting faster. Tape formats are storing more information on less tape and getting it back more reliably.
Ways of protecting data are working over longer distances, between ever shorter intervals, and making restoration easier, faster, better.
Yes, there is acronym overload, concept confusion, sale by FUD ever listened to a compliance pitch? How do you like buying a product because the selling pitch threatens you with jail?
But thats just marketing, sheer sales fizz. Under it all is human ingenuity at work.
- FC SANs too complex? Lets try Ethernet ones,
- Not enough I/Os from 3.5 inch disks? Lets have smaller ones,
- Parallel interfaces getting limited. Then lets try serialised ones.
Its what we people do. We restless developers, manufacturers, suppliers, managers and users of technology. We want to make things better.
One area where storage development is going to have a lot of tensions is in the iSCSI one. An iSCSI SAN will be managed, will be virtualised, by an intelligent controller at its head. There is no Ethernet infrastructure equivalent to Fibre Channels switches and directors which take some responsibility for supplying the storage resource that passes through them.
Indeed, the latest development is that fabric directors are intelligent platforms for storage array virtualisation and layered applications such as backup to virtual tape or migration or replication.
This just cannot be done in the Ethernet world because there are no equivalent boxes in the LAN.
It means that FC SAN management software will be different from iSCSI SAN management software. Which in turn means that FC SAN-owning enterprises, wary of having SAN islands, unwilling to have two different SAN management products and processes and disciplines, wont buy iSCSI SANs in any significant way at all.
Which means that iSCSI SANS are for small and medium enterprises.
How hard is it to combine NAS with an IP SAN? Surely its easy. All you need do is add software to the IP SAN controller box which will probably be either Windows or Linux-based. Take an existing IP SAN product and bolt it on.
And whats to stop any NAS vendor, locked inside NAS commodity hell, try to break out by adding an IP SAN behind their NAS box? Imagine the telephone call . Hello, is that DataCore? Ive got a NAS product based on a Windows Intel server box. About licensing SANmelody .
You see how easy it might be?
Such a combined NAS/IP SAN would have one virtualisation product and one management product.
Businesses that buy iSCSI SANs and get NAS functions too are just not going to buy Fibre Channel unless they have a really good reason. Why should they? FC represents added complexity, added management overhead, added cost.
In fact, for them, the letters FC stand for Fat Chance.
We have a serious chance of a SAN schism developing, which will place strains on vendors trying to supply products to both Fibre Channel SAN and IP SAN customers. They are going to need two different sets of products. So much for the single storage utility.
Storage problems are becoming bigger
If you think we have it bad in terms of the data storage burden today then the good news is that you do, youre right. You really do have a tough job.
The bad news is that its going to get worse. Imagine the combination of compliance regulations, specifying retention of digital data, and the entry of voice telephony into the digital data sphere.
VOIP is just a stream of bits, and compliance regulators make us store bits like a whole flock of demented squirrels burying nut mountains under our lawns.
RFID tracking of things components, products, even people is going to bring a whole new world of bits to reside on your spinning platters and streaming tapes. You can see the drive and tape suppliers eager to shake their heads and sympathise;
What an awful burden this is going to be.
Here, have another hundred Barracuda drives.
Gosh, those regulators, that product tracking, its a dreadful world isnt it?
Crocodile tears, every one.
Branch office consolidation will pour even more data into the datacentre silos.
Mount Everest might have lost a few feet recently but our data mountains are growing incredibly fast, squeezed ever upwards by a new set of tectonic plates mashing together VOIP, RFID, compliance and branch office consolidation.
Our data currencies today are gigabytes and terabytes. What about megabytes? Thats RAM. And even thats moving to GB measures. We dont use MB for measuring disk any more. Its like, so yesterday!
What about kilobytes? That unit isnt even used for cache.
We are transitioning even as we have this dialogue, this show, to a world of terabytes and petabytes.
Keep your options open
If you choose the wrong technologies then you could enter a storage blind alley.
- Maybe locked in a fibre channel world when iSCSI is the way to go,
- Possibly prevented from having the benefits of a fabric storage intelligence and management platform because you have gone 100 percent Ethernet,
- Perhaps having a single vendor for integration reasons but losing out on other technology
Our storage world is full of siren voices, luring you towards their technologies, their products, their services.
Which are the right siren voices for you? You cant use all the technology out there. You need to try and strike a balance between what you do need, what you cant do without, on the one hand, and keeping your options open on the other.
Because next year the technologies will be better, the products will have moved on, and you dont want to be locked into a storage blind alley. Reversing a technology or product direction is a hard thing to do.
Try to detect the mainstream technologies that suit you.
Perhaps you do need continuous data protection, but perhaps a faster intermittent protection will do. Is sync replication mandated by your needs? Will async suffice? Can SATA drive arrays really replace a portion of your tape archive?
Storage is terrific stuff. It's developing in myriad new ways. Some of these technology shoots will become cul de sacs, blind alleys. Do your best to avoid them by keeping your options open, choosing mainstream technologies in general, and only go for a niche technology if it answers a real need you have in a clearly better way.