Have you tried rebooting?

How does cloud affect the storage infrastructure?

// November 16th, 2011 // No Comments » // Technical Know-It-All

Surely everywhere you venture these days, there is talk about Cloud. A year or two ago, it was limited to the Enterprise, but now it seems that its found its way onto the consumer space as well (eg; iCloud).

I’m not the foremost expert in cloud, but I think everyone’s got a definition for it today. You decide if what I’m saying makes sense.

Is cloud a product?
I don’t think so. Its almost like asking if Internet is something you can buy.

Is cloud a technology?
Hmm… in some ways, but not completely accurate.

My definition is really that Cloud is more of a CONCEPT (for providers) and SERVICE (for users).

Think about it for a second, all you ever hear about cloud is really about a service you can buy or provide. Example being, IaaS (Infrastructure as a Service), PaaS (Platform as a Service), SaaS (Software as a Service) and etc. So don’t be completely fooled when vendors trys to sell you Cloud…

I think in-terms of storage, what vendors are really selling would probably be better termed “Cloud-Enabled“.

Features such as thin provisioning, wide striping, data resiliency, storage virtualisation, dynamic tiering and charge backs tools are some of the key features that makes sense to a cloud deployment. How so? I’m not gonna deep dive into the details but hope you will get the drift…

If you ask me, Cloud is just another buzz word in recent times that describes the evolution of the IT industry… Lets take a stroll back memory lane…

1. First there were built-in hard-disks. We realised that utilisation was low, so we built SAN’s for “sharing” storage.

2. We built the PC, but we wanted to communicate with the next PC. So we hooked them up. We realised then, we needed to “share” data with more PC’s, thus we built the LAN, then later the WAN, then the Internet.

3. Servers were great functioning as physical boxes. Again, low utilisation meant that we wanted to “share” to better utilise the resources. Tada… Server virtualisation!

4. Now we have SAN’s, NAS’s and etc. We realise that every vendor talks their own language. EMC doesn’t link with HDS and vice versa. So we have pockets of storage that is not properly utilised. If you don’t already know, this is now happening, and its called Storage Virtualisation, where all storages are consolidated into a single pool regardless of vendor. So we can “share” more efficiently.

5. What’s next on the roadmap? Not too sure if you have heard, FCoE and CEE is around the corner. In a nutshell, that is again, consolidation of SAN and LAN, so the networks can “share” a single common infrastructure.

So, if you look at the 5 points above, the common word amongst them is really sharing (in the enterprise, it’s better known as “cost saving” ).

Looking at cloud again, we combine all of the above, and again look at the possibility of further reducing costs and/or sharing. And the Cloud is born. Use the existing infrastructure in its entirety and share it out at a higher level.

So when shopping for your next enterprise purchase, be it storage, software or servers, make wise decisions as to how its features can be shared. Do not just take the vendors word for it, cause some just don’t quite make sense.

SAS vs FC Disks

// November 8th, 2010 // No Comments » // Technical Know-It-All

In recent times, there has been much said about the future of SAS (serial attached scsi) vs FC (fiber channel). Don’t get me, I’m not saying that FC is dropping dead tomorrow. It is probably still relevant for another 3-5 years, but I for see the future in SAS disks.

Here’s my quick lowdown on it. In no way conclusive, but its something to ponder about.

1. Why SAS?

Why the crap not? SAS is a fraction of the price cheaper, has a strong roadmap moving forward with 6Gbps today, and moving ahead in the future to faster backends. FC has been at 4Gbps for the longest time and have not moved an inch since. It also doesn’t have a roadmap to go any further beyond 4Gbps.

SAS also comes in 2.5″ variants. Have a real estate or green issue? Here’s to a smaller, lighter and greener media.

2. But SAS is Tier 2 disks

How do we define Tier 1 from Tier 2 disks today? Reliability and performance.

In terms of reliability, SAS disks are manufactured and developed to the same specifications of a FC Disk. The only difference is the connector / interfaces at the backend. The likelyhood of a SAS drive failing is the same as a FC drive failing.

Performance, used to be a big discussion point because SAS ran at 3Gbps, but now with 6Gbps and a non-arbitrated loop architecture, it is blowing FC off it’s socks. No discussion there.

3. SAS is SATA’s expensive cousin

Very good observations there. True! The commonality here is “Serial”, and because of that, you can now intermix SAS and SATA disks in the same enclosures. No need for different enclosures for different disk types and no more for funky FATA disks. Go forth and intermix. Beyond that there is nothing similar with regards to the built of the drives.

4. Not convinced still?

Surely vendors like Hitachi Data System’s (HDS) or even IBM can’t be so far off their socks. Hitachi for example have a Hard Disk’s Division that only builds drives. Their business is far far larger than their HDS counterparts. So do you think they will just drop their FC product roadmap just to help out a smaller subsidiary of theirs? Or do you think they are considering the shifting market trend?

Ultimately, there is always the opinionated Google that speaks for itself.

Free SAN Technical EBooks

// March 23rd, 2010 // No Comments » // Technical Know-It-All

I came across 2 ebooks that were free recently. Not too many things are free these days but I must say these 2 books are pretty good entry level books. Enjoy!

Brocade CLI Zoning Tutorial

// March 11th, 2010 // No Comments » // Technical Know-It-All

Given that more and more features within Brocade SAN switches are configured using zones, thought I would share this quick CLI guide for those looking to script this.

Zoning is organised as such

- Every SAN device has a unique WWN (much like how IP devices have MAC addresses)
- WWN’s are then mapped to aliases
- Multiple aliases then form a zone
- Multiple zones then form a config
- Nutshell, WWN->Aliases->Zones->Config

So here are the steps to create a zones quickly using CLI.

1. Ensure you have noted down the device WWN and assign it a usable name for alias. WWN’s for devices can be found using “switchshow“. In this example, I will be mapping the following devices to the appropriate names

10:00:00:00:00:00:00:01 -> Server_HBA01
50:00:00:00:00:00:00:01 -> Storage_P01
50:00:00:00:00:00:00:02 -> Storage_P02

alicreate “Server_HBA01″, “10:00:00:00:00:00:00:01″

alicreate “Storage_P01″, “50:00:00:00:00:00:00:01″

alicreate “Storage_P02″, “50:00:00:00:00:00:00:02″

2. Next, we want to group these 3 members into a single zone so they can have access to each other. Let’s create a zone called “Zone_svr2_hba01_str_p01_str_po2″. Remember that zone members are case sensitive.

zonecreate “Zone_svr2_hba01_str_p01_str_p02″, “Server_HBA01; Storage_P01; Storage_P02″

3. After completing step 2, put the newly created zone into a config. For simplicity, I shall call this “Config_Test”. Again this is case sensitive.

cfgcreate “Config_Test”, “Zone_svr2_hba01_str_p01_str_p02″

4. Verify the zones using “cfgshow“. The zones should only show up under “Defined Configuration”. To make this zones active, you will need to enable it.

cfgenable “Config_Test”

5. Now if you run “cfgshow” again, you will realise that the “Defined Configuration” and “Effective Configuration” is the same. This is when you know the zones are activated.

6. To script this, you can simply prepare all commands in Notepad and open up a putty session then cut and paste it. You can also use SSH scripts if required.

The Buzz About FCoE and CEE

// February 26th, 2010 // No Comments » // Technical Know-It-All

The new buzzword these days in the Storage / Networking space is FCoE (fiber channel over ethernet) and CEE (converged enhanced ethernet). For those who are unfamiliar with the whole concept, in layman terms, it is a convergence of the traditional TCP/IP networks and traditional SAN networks into a single giant network.

While it is ideal from a C-Level perspective to streamline processes, reduce costs, go green and try doing more with less, at the technical level, it is never as easy as the vendors make it seem. Many industry experts have claimed that this new protocol will eventually replace the ever resilient FC Protocol and Ethernet. The jist of it is that, everything that has to do with “networking” will run on this newly improved protocol.

While this “could” be true, here are some points to ponder upon before taking the leap of faith to run your entire production systems on CEE and FCoE.

1. TCP/IP has never been known to be the most secure protocol (before I get flamed by the networking experts, it can potentially be secured with the right amount of tweaking). Compared to FC, while some may disagree, it is more secure in some ways (for starters, given that not every laptop will have an FC port to begin with, but having said that, more often than not SAN administrators take this for granted). By merging the 2 protocols together, suddenly we have a scenario where security might now be a concern to the traditional SAN guys.

2. Usually in large environments there is usually a Storage Team and Network Team. With the convergence, a discussion of job roles would probably emerge, and potentially job cuts (fingers crossed).

3. Standards. Currently, many of the protocols/standard are still very much in the review stage with IETF, T11 and the lot. Ethernet is stable today after many years of revision while CEE will take some time before reaching that level of maturity.

4. Physical infrastructures such as cabling are probably not ready today for a complete deployment. Many are still running cabling works that are barely rated for 1Gbit, let alone the 10-40Gbit as promised by CEE/FCoE. Also, existing Ethernet switches and FC switches are not meant to run this new protocol. Slowly and surely, vendors are releasing firmware updates to enable them to support the protocol in the legacy units, but having said that, the additional overheads of the protocol was never meant to run on existing hardware, so we might never know if it will be able to handle it. Newly released FCoE/CEE ready devices are slowly coming to market and today cost quite a fair bit.

Obviously, my little list above is not all conclusive, but my 2 cents of what I would probably be concerned about when thinking of rolling it out.

So in summary, I am not saying that CEE/FCoE is doomed for failure but maybe, just maybe, it is still too early for it to be totally viable in a full fledge production environment. I do believe that CEE/FCoE will take off and be a prominent protocol some day. Having said that I don’t think it will oust or replace traditional FC and Ethernet anytime soon as some major vendors are suggesting. It takes more than lowering cost and bandwidth to achieve total protocol dominance.

Just ask iSCSI and Infiniband.