Archive for the ‘Storage’ Category

EMC Symmetrix File System (SFS)

March 8th, 2010 4 comments

Very little is known about the Symmetrix File System largely known as SFS. Symmetrix File System is an EMC IP and practically only used within the Symmetrix environment for housekeeping, security, access control, stats collection, performance data, algorithm selection, etc.

If there are any facts about SFS that are known to you, please feel free to leave a comment. This post talks about the effects of SFS and not really the underlying file system architecture.

Some facts about the Symmetrix File System are highlighted below.

  • Symmetrix File System (SFS) resides on volumes that have specially been created for this purpose on the Symmetrix
  • SFS volumes are created during the initial Enginuity Operating Environment load (Initial install)
  • 4 Volumes (2 Mirrored Pairs) are created during this process
  • SFS volumes were introduced with Symmetrix Series 8000, Enginuity 5567 and 5568


  • 4 SFS volumes are spread across multiple Disk Directors (Backend Ports) for redundancy
  • SFS volumes are considered as reserved space and not available to use by the host
  • Symmetrix 8000 Series: 4 SFS volumes, 3GB each (cylinder size 6140). Reserved space is 3GB x 4 vols = 12 GB total
  • Symmetrix DMX/DMX-2: 4 SFS volumes, 3GB each (cylinder size 6140). Reserved space is 3GB x 4 vols = 12 GB total
  • Symmetrix DMX-3/DMX-4: 4 SFS volumes, 6GB each (cylinder size 6140). Reserved space is 6GB x 4 vols = 24 GB total, (It’s different how the GB is calculated based on cylinder size on a DMX/DMX-2 vs a DMX-3/DMX-4)
  • Symmetrix V-Max: 4 SFS volumes, 16GB each, Reserved space is 16GB x 4 vols = 64GB total
  • SFS volumes cannot reside on EFD (Enterprise Flash Drives)
  • SFS volumes cannot be moved using FAST v1 and/or FAST v2
  • SFS volumes cannot be moved using Symmetrix Optimizer
  • SFS volumes cannot reside on Vault Drives or Save Volumes
  • SFS volumes are specific to a Symmetrix (Serial Number) and do not need migration
  • SFS volumes are managed through Disk Directors (Backend Ports) only
  • SFS volumes cannot be mapped to Fiber Directors (now FE – Frontend Ports)


  • SFS volumes are write enabled but can only be interfaced and managed through the Disk directors (Backend Ports).
  • SFS volumes can go write disabled, which could cause issues around VCMDB. VCMDB issues can cause host path (HBA) and disk access issues.
  • SFS volume corruption can cause hosts to lose access to disk volumes.
  • If SFS volumes get un-mounted on a Fiber Director (Frontend Port), can result into DU (Data Unavailable) situations.


  • Since the SFS volumes are only interfaced through the Disk Directors (Backend Ports), the PSE lab will need to be involved in fixing any issues.
  • SFS volumes can be VTOC’ed (formatted) and some key information below will need to be restored upon completion. Again this function can only be performed by PSE lab.
  • SFS volumes can be formatted while the Symmetrix is running, but in a SCSI-3 PGR reservation environment it will cause a cluster outage and/or a split brain.
  • No Symmetrix software (Timefinder, SYMCLI, ECC, etc) will be able to interface the system while the SFS volumes are being formatted.
  • The security auditing / access control feature is disabled during the format of SFS volumes, causing any Symmetrix internal or external software to stop functioning.
  • Access Control Database and SRDF host components / group settings will need to be restored after the SFS format

Access / Use case

  • Any BIN file changes to map SFS volumes to host will fail.
  • SFS volumes cannot be managed through SYMCLI or the Service Processor without PSE help.
  • SYMAPI (infrastructure) works along with SYMMWIN and SFS volumes to obtain locks, etc during any SYMCLI / SYMMWIN / ECC activity (eg. Bin Changes).
  • Since FAST v1 and FAST v2 reside as a policy engine outside the Symmetrix, it uses the underlying SFS volumes for changes (locks, etc).
  • Performance data relating to FAST would be collected within the SFS volumes, which FAST policy engine uses to gauge performance.
  • Performance data relating to Symmetrix Optimizer would be collected within the SFS volumes, which Optimizer uses to gauge performance.
  • Other performance data collected for the DMSP (Dynamic Mirror Service Policy).
  • All Audit logs, security logs, access control database, ACL’s etc is all stored within the SFS volumes.
  • All SYMCLI, SYMAPI, Solutions enabler, host, interface, devices, access control related data is gathered on the SFS volumes.
  • With the DMX-4 and the V-Max, all service process access, service processor initiated actions, denied attempts; RSA logs, etc are all stored on SFS volumes.


  • SFS structure is unknown
  • SFS architecture is unknown
  • SFS garbage collection  and discard policy is unknown
  • SFS records stored, indexing, etc is unknown
  • SFS inode structures, function calls, security settings, etc is unknown

As more information gets available, I will try to update this post. Hope this is useful with your research on SFS volumes…



HP Blades Day 2010: Paul Perez [CTO, Storageworks]

March 8th, 2010 No comments

During HP Blades Day 2010, we had a 30 min chat with Paul Perez – HP CTO, Storageworks Division.

Some of the highlights of the discussion with Paul are below

  • One master provisioning stack for all Storage
  • VDI solution from HP
  • Integration of IBRIX and Lefthand – Common Management Infrastructure
  • Proliant features of power management etc, to be passed onto the IBRIX and Lefthand with Convergence
  • Customer References
  • Bottlenecks in networking and storage today
  • Stranded pools of IP (Intellectual Property) within HP, Integration challenges
  • DataDomain
  • Integration of HP EVA teams with HP Unified Storage Division
  • Storage Tiering
  • “Memristor” – HP Labs
  • Industry Consolidation, business models
  • Convergence is Next Gen

The Video – Paul Perez Unplugged

Part 1

Part 2

Part 3

HP Blades Day 2010: Final Thoughts

March 3rd, 2010 1 comment

This is my 5th consecutive post on HP Blades day.  So far no videos have been uploaded, just the coverage of the event and pictures. This post primarily focuses on what I feel we saw at HP in terms of things that will help them, challenges in the market and where all this may go.

There has been at least 6 to 8 hours of video recording on my flip camera; starting tomorrow I will upload these videos only of the most interesting sessions on the blog.

Satellite View of HP Facilities in Houston, TX

The coverage of the event can also be found on Greg Knieriemen’s Infosmack Podcast on Storage Monkeys, here.



This event was a very smart move by HP and as far as I can see they have exceeded their expectations with this event. Though I felt the twitter activity with HP Tech Day (Storage, #hptechday) was much higher than what we saw with this event HP Blades Day (Blades, #hpbladesday). Though the after discussions have taken over the blogging, twitter and the Internet press by surprise with the number of tweets, blogs and press articles written about this event.

Clearly for me this was a good platform to learn, understand and share some visions and technologies related to HP Blade products. I have been a storage focused individual, but only with a shallow knowledge of the blades architecture and infrastructure. This was a great event for myself to understand the depth of these products and take a deep dive into the interworking of converged infrastructure. An Event like this helps understand and connect the dots together with future products and emerging technologies. As this was a non-NDA event, we didn’t have preview to the next generation of HP Blade products.

One thing that is pretty visible and positive is that HP has managed to mobilize resources in the direction of integrating internal resources relating to converged infrastructure. Though its obvious and again visible that at places, they have not been able to fulfill that dream entirely.

There were some awkward moments where the engineering teams were asked to not do a deep-dive on other vendor technologies. The marketing folks spoke about some strategy related to these technologies and painted an overall picture. The mix of people involved with the presentations and demos seem to accomplish the agenda. Marketing pitches by social media and marketing teams along with engineering details by the architecture teams seemed to accomplish their goals.

The highlight of the sessions were a 45 min talk with the CTO of StorageWorks, Paul Perez and the competitive intelligence session that was hosted my Gary Thome and his team to compare HP Blades products with Dell, IBM and Cisco UCS. Discussions around CEE and Virtual Connect were pretty interesting.

HP emphasized the 250 million dollar investment with Microsoft over and over during the HP Blades Day. This proves that they value this partnership heavily and possibly have a roadmap associated in the future with great integration with Microsoft products.



HP emphasizes a lot on converged datacenters and the products it’s gearing for the next generation. But an integration vision from a convergence management was still lacking, a direction or a strategy on how these pieces of puzzle will be joined together and managed. HP clearly owns all the stacks of the next generation products, but again the orchestration and integration is one thing that is not very clear yet. Say too big and too much to manage!!!

With Networking products and the focus on Virtual Connect, HP seems to be moving in the right direction, but again when it comes to FCoE and CEE (Converged Enhanced Ethernet) the direction is pretty unknown. It’s sort of wait and watch as to where the market goes and drives demands, a lack of vision in real terms. HP clearly has a big competition from Cisco when it comes to the Networking stack.

With Storage products, HP clearly has a very big competition with already proven Vendors and their technologies like EMC, NetApp and IBM. Also technologies that are strong and emerging would largely cause market nuisance or focus disruption for HP.

With the Blade products, HP is a market leader, but truly considers Dell, IBM and Cisco as the biggest threats and sort of prepared to fight against it. Seems the next generation Rack and Blade products might seem to have a lot of integration with storage and networking.

The services story, with the acquisition of EDS, HP made a move in the right direction being the first in the market to do so. With the latest acquisitions from Dell of Perot Systems, from Oracle of SUN Microsystems and by Xerox of ACS, large vendors are all trying to fulfill the services gap. HP clearly has a big competition with IBM and Oracle in the space.

The VCE (VMware, Cisco, EMC) coalition: What are your thoughts. It’s pretty amazing to see HP not mention the word ‘cloud’ these two days. Focus has been virtualization and the partnership with VMware, but really no focus on moving toward utility market and integration of all next gen products for converged datacenters with the underlying virtualization layer. May be the Microsoft partnership may fulfill this.

VMware or Microsoft: They didn’t say this, but seems something is cooking. The partnership with Microsoft and the investment of 250 million dollars will create some friction with VMware, at least my guess. Next gen products may utilize Hyper-V as an underlying virtualization layer rather than using the default VMware Hypervisor.

HP still needs a very strong storage technology in the Enterprise space that is their own and not OEM’d. The truth is, eventually the HP – Hitachi relationship has to come to an end with HP’s new product that may compete in the same market space. This strategy will enable HP to be very unique in terms of the markets they serve, which may include their own in-house storage products for SMB, Midsize and Enterprise customers.

So other lacking things from HP were the Cloud Strategy (if they ever plan to enter that space), FCoE discussions, Procurve and Storage Management as it relates to Insight Software.

It may have been very hard to cover all these platforms in a day and a half with giving all the technology details behind it. Also remember this was a non-NDA session, so we were not preview to all the future products and technologies.



Overall HP did hammer us for 2 consecutive days with HP Blades Technology. Coming out of it, I can truly say, HP had so much focus on datacenter convergence. Their move to hire Dave Donatelli was a smart one many of his strategic moves and direction in the ESSN (Enterprise Servers, Storage and Networking) are pretty visible now.

Apart from GestaltIT Techfield Day, HP is still the only OEM to arrange Bloggers Invite Only Event. The ratio of Bloggers to HP Personnel was 1:2, giving everyone a lot of attention.

Now the question is who will be next OEM to do a similar event and what will they do to prove themselves different. Already hearing some buzz in the industry about some the effects of HP Blades Day and some possible events from other OEMs.

But I clearly see an advantage of an event like this and the after effects of it, good move HP Marketing Team! Along with Ivy Worldwide!!


Disclaimer: This event is sponsored by HP and hosted in Houston, TX. HP paid all the flight, living and mostly food expenses. This is a bloggers – invitation only event. No products have been given by HP.

HP Blades Day 2010: Friday – Day 2

March 2nd, 2010 No comments

Again an early AM start today, with a busy schedule ahead at HPBladesday.

The day kicked off with some Q&A on the topics covered from the previous day. We all had full access to the team from yesterday to see if there were any questions.

Most of the questions were around FCoE, Convergence, Blades infrastructure and the discussion on the Tolly Report.

Tolly Report

Some discussions on this report by Stephen Foskett

Some discussions on this report by Kevin Houston

After the Q&A session, Joseph George took the stage with a presentation on Client Virtualization and HP product overview; covering the VDI solutions offered my HP. This was an insightful presentation on how the HP blades and software infrastructure can be used with VDI.

Joseph George

Kevin Houston, Rich Brambley, Simon Seagrave

Martin Macleod and the HP Team

After a demo of the Client Virtualization lab, we went for the Factory Express tour.

Chris Evans

The factory express tour was presented my Denise Herdman, Jean Brandau and Wade Vinson. We all got to see how the HP systems were assembled both the Servers and Storage in different factories. How they performed quality testing, how they were all packaged, etc. This was a great tour of the facilities giving us a full access to how supply chain works within HP from a customer entering an order to how the equipment is fulfilled. Quite an operation….

Greg -

After the factory express, we were given the HP pod (performance optimized datacenter) tour; think of this as a datacenter in a trailer. Wow…. it was impressive. Here are some pictures. These HP pod datacenters are 40% more efficient than the typical conventional datacenters.

Simon Seagrave in the POD

Systems inside a POD

The day was coming to an end. We were at the last destination of the day, at a Mexican restaurant having lunch. After the lunch headed back to the hotel. It was time for the final goodbyes to the Bloggers, HP and Ivy worldwide teams. But there were about 5 of us staying back that day.

Enjoyed the rest of the day, a bit of a relax, worked out at the gym, had a quite dinner, a drink and hitting the bed early to leave back home next morning.

Though both the blog posts from today and yesterday were just an overview of the activity we did at HP, some deepdive sessions and videos from the event are forthcoming.

Disclaimer: This event is sponsored by HP and hosted in Houston, TX. The flight, living and mostly food expenses were all paid by HP. This is a bloggers – invitation only event. No products have been given by HP.