Storage Most Recent Posts

After the initial R2 hoopla, and in the context of storage, Win2k3 R2 really has some interesting features that seem to me to indicate some of Microsoft's struggle to maintain their engagement of the marketplace with the traditional Windows filer approach.  The more I look at the features in R2, the more I think there's some obvious statement Microsoft is making about competitive pressure in this area.

I think R2 in a lot of ways was a major disappointment to folks because of features gradually removed on the way through beta.  In my opinion, some of the best features remaining in R2 are the out of the box filer optimization associated with Storage Services, and also the Single Instance Storage (SIS), so why are those features there, out of everything initially promised in R2?  Those features seem to say a little bit about MS's position on storage.  The further work on DFS might be worth a bit, but DFS still seems to fall a bit short of the mark for many of the needs it seems geared toward (think branch office replication).

Are you finding your email storage and archiving expenses climbing through the roof? If you are an email service provider, you know that every subscriber is guaranteed a certain amount of disk space to store messages. If you are in enterprise operations, then you know government regulations mandate message retention for approximately seven years.

One of the easy to measure costs of spam is the expense of storing all of that junk mail. The less effective the spam filter, the more you pay for storing the spam that gets through – nice double whammy! Cloudmark (www.cloudmark.com), one of the leading anti-spam vendors, suggests that approximately 60% of stored messages are spam. Spam messages tend to be smaller since they seldom drag along big attachments, so the vendor estimates that over 20% of your email storage requirements are consumed by spam. That means that one out of five messages stored and managed is useless spam. At well over $10,000 for a terabyte of storage, the operating expenses go up very quickly.

SAN design has some pretty straightforward basics for most mid-sized environments where you're not dealing with a lot of ports, even while a variety of switch and host choices and configurations exist.  But as we're executing a storage implementation project, we have a real need today to identify long term expansion strategies, and a view into a much larger environment would assist our design and growth projection activities today. 

So Cisco, where's my blueprint for SAN design?

I haven't found it yet if it's out there, maybe someone can point me to it if I'm missing it.  I know the design expertise is there, and a webcast by Dan Hershey from Cisco over on Search Storage has some excellent overview material (Look for the session titled "How to Design the Fastest, Most Scaleable Enterprise SAN Architecture").  Maybe the best captured overview of the MDS and SAN design I've seen.  Now where's my blueprint Dan?

Looking through my posts here probably makes anyone looking wonder if I have an obsession with Virtual Tape Libraries. Ultimately I think the market is just emerging and I can't find my perfect product. There seems to be some real disparity between vendors in the solution approaches they offer, although I do think they'll largely converge over the next year or so. So here it is, my wishlist for VTL. While I know this wishlist is unrealistic, this would be an ideal solution.

An ideal VTL

1. Present the commonly seen VTL solution front end which emulates tape drives of the user's choice.
...Read more

Cisco announced a series of upgrades to their MDS line of storage switches at Storage Networking World in San Diego last week.  Maximum port counts were upgraded to a whopping 528 ports per director; 4 and 10 Gbps support was added across the entire MDS product lineup; and a new larger director chassis was introduced.

In my opinion the bottom line with this announcement is two-fold:

1. Cisco has the biggest director on the planet from a port count perspective

Cisco now tops out at 528 ports per director compared to 256 ports for McData and Brocade.   Brocade and McData argue that you can't use all of those ports at full speed at the same time which is true - only 132 ports can be run simultaneously at 4 Gbps.  I contend that this "over-subscription"  issue is a non-starter for most data center managers.   Fibre Channel networks are typically configured in an over-subscribed manner. Each port into a shared storage system is typically shared by many servers - up to eight or more is common.  Yeah, full bandwidth into and out of all ports in a director is architecturally more predictable, but given the low utilization rates on most FC networks, and the trends towards scaled-out clustered architectures and server consolidation, I believe Cisco with more ports per director is solving the more pressing need  - more connectivity.  In other words, I believe that for the majority of users, more ports in a singly managed platform is more desirable than guaranteed line rate on every port.

Last week, several hundred IT folks headed to San Diego to see the latest and greatest in storage at SNW (Storage Networking World).  While many show-goers listened to pitches about next generation storage systems and a variety of new methods to protect data, this SNW was the first event where information classification technology made its way from 'whisper suites' to the trade show floor.  Information Classification, the process of preparing data for action, is a critical market segment within Intelligent Information Management helping organizations contextually understand their information assets.

I'll admit, I'm a bit removed from Information Lifecycle Management (ILM) concerns - I have enough problems upfront just addressing capacity and infrastructure, and ILM is a couple of steps removed. But with that in mind, I keep trying to get my finger on the pulse of the market to lock down some plans for a longer term strategy. From my perspective, it seems like we've ended up in a data hole with no good way to dig ourselves out.

I'm about as tired of hearing the ILM acronym as the next guy, and on the surface the need is real, and the solution seems simple. You got a bunch of stuff right? Step 1, figure out what you've got, step 2, take some action on it. Hey, we're used to moving data around all the time, right?

I've seen a lot of interesting file services solutions here at SNW, and they've ended up making me really question our overall approach to file services and how we got here.

I ran into the neopath solution for the first time at SNW. Cool product, but it pretty much creates a redundant service in your data path, by managing multiple CIFS/NFS servers behind a pool of CIFS/NFS neopath heads. The product allows you to do a bunch of stuff, but relevant to me is some pretty neat data migration and HSM type stuff, if you have multiple CIFS servers attached to multiple performance or capacity levels of storage. But I'm struggling with adding another level of CIFS servers, and still having to manage wintel infrastructure behind it. Not to mention actually needing to add more wintel servers attached to different levels of storage. So for me, I have to question whether it's worth the complexity, and question what the roadmap when the product changes or disappears over time.

So at the conference Microsoft announced they will have an iSCSI target component which will be released in the summer of 2006. They purchased the software from String Bean which was called WinTarget. Microsoft's press release on the String Bean acquisition can be found here.

My eyebrow notched up a few millimeters when I noticed the brand-name of my hotel soap was "Archive." Sting's got nothing on this conference. But archival is definitely one of the reasons I am at Storage Networking World.

Interestingly however, the topic I keep running into here is security. Encrypting the disk, the link, authentication... the CISSP folks, and others, definitely have their ears to the storage railway. That's good news, no doubt. Especially when I think iSCSI. The attack surface of iSCSI is huge, so long as there are iSCSI targets and initiators peppering the internet

So the LSI guys look like they rock. And that's hard to do for a company that on the surface makes cards and components (although there's a lot more to them if you look into it further). I'll admit to being a bit of a hardware geek at heart, if I can touch it, feel it, overclock it, whatever, I like it a little better. So some of my enthusiasm might be a little biased. LSI is here with two products that speak to the hardware geek in me.

First, LSI Logic is doing some nice looking SAS drive controllers, with what was previously SATA/IDE raid feature sets in a controller card probably appropriate for small server / workstation use. Similar to what you find in the Promise and 3Ware type controller cards. Not bad, I wasn't aware anyone had product here yet, and it's nice to see the development happening. Looks like these can take about 10 devices and support a variety of striping/mirroring or a system that LSI calls RAID 1E/10E which is a funky kind of interleaved striping to allow the controller to support some protection on odd numbers of disks while still seeing benefits of striping. The explanation was enough to make me have to think, so I'll have to run through the description in my mind a bit more before it makes perfect sense (sounded like shifting the mirror of each stripe to the next disk, i.e. given N disk stripe, N disk stripe's mirror is on N+1 disk).

I spent a bit of time talking with Avamar this a.m.. In an earlier post, I mentioned that I actually deferred making a VTL/D2D commitment from this year's budget cycle and elected to acquire other functionality instead, as I couldn't get comfortable with a VTL/D2D solution set. Avamar Axion seems to have some real promise here, and I suspect I'll be talking with them further in the future. They offer the best combination of a couple of key factors that are important to me:

1. They are software only, and can run on any server hardware or storage. (For the record, our 3.5TB of file services that would reap primary benefit from this are on the Windows platform, so I haven't looked into multi-platform support)
...Read more

Content aware storage (CAS) is one of the fastest growing areas of the storage market.  Companies are implementing CAS solutions to meet the needs of compliance regulations, litigation protection and to enable better IT efficiencies.  CAS storage systems set up retention policies on objects (file + metadata) preventing anyone from editing or deleting them for a period of time.  In some cases the period of time can be three years and in other cases it is forever.  The method for creating retention periods is proprietary from system to system.  This creates a major issue for companies that are implementing CAS systems - vendor lock-in.

So I was surprised by Rob Pelgar's sessions on virtualization and how good they were, and pleased with the first grid session from HP, but it seems like the Hitachi Data guy must be doing the second grid session, and it's more principles and practices oriented and I'm not real engaged. I was hoping more for future directions and how they might directly impact the enterprise, as some talk in this area was really what pushed me over toward the HP EVA solution and some of the complications inherent in running multiple storage chassis/controllers and the SAN islands they tend to operationally create. HP has some ambitions I think for virtualizing their storage more seemlessly across controllers as a part of their grid computing initiative. I think that's a key area in which grids can deliver value to the enterprise.

So, as a further update on SNW 06 in San Diego, some of the best sessions I've seen so far have been Xiotech's Rob Peglar's product agnostic SNIA sessions on virtualization and the following sessions from HP's Abbot Schindler on Storage Grids which Rob himself seemed pretty keen on. In fact, as I sit here in the session Rob is sitting in the chair right in front of me attentively listening to Abbot's second session on grids (in fact, I bet I'm bugging him with my keyboard chatter). These sessions together have raised some more interesting questions on where in the network virtualization and management intelligence should be.