Internal Clouds – Marketing the Non-Existent

I was recently asked my opinion on what were the main considerations for Cloud Computing with specific emphasis on Internal Clouds. Eager to assist I quickly gave a rundown of issues which included SLAs, distinguishing charge rates, security etc, etc. Pleased with the response received our conversation then veered off into another direction but then it struck me – I had just fallen victim to the marketing jargon. Internal Cloud? What on earth was he asking me and what on earth had I just answered with?

I thought back and reassessed my understanding of the Cloud to what I originally understood it as i.e. that the Cloud was beyond the realms of the datacenter and certainly not internal. Facts told me that Cloud storage services whether it be a backup or archive reside outside of the local datacenter and into the ‘clouds’ even to the extent of being in another continent.

So how then does the oxymoron of ‘internal cloud’ exist so much so that in depth conversations are taking place between consultants, IT managers and Storage Engineers at a datacenter near you? The answer is simple; marketing. Not content with pushing Cloud and SAAS as the future low end tiered storage, the term ‘internal clouds’ is now being marketed and ascribed to new and upcoming products which in essence are only offering added support for virtualization or clustering.

The metaphor of an ‘internal cloud’ i.e. a cloud which now exists within the datacenter leads to a rather ironic image of a misty atmosphere that causes even more confusion. Blow those ‘internal clouds’ away from your datacenter and what you’ll see are flexible solutions for scalability whether they are in the guise of global namespaces, clustering, grid storage or virtualization; solutions which were already known about and quite possibly already existed within your datacenter but were now coined as ‘internal clouds’. Hence once the haziness has disappeared it’s clear to see that the internal cloud that we’ve been marketed with never really existed in the first place.

So should I be asked my opinion on internal clouds again, let’s just say that this time my answer will require no words but a straightforward raise of the eyebrow and a protruding of the chin.

'NetApp & Microsoft' vs 'EMC Cisco VMware'

2009 seems to have been the year of corporate marriages but whether they are marriages made in heaven is yet to be seen. The ongoing saga between Oracle and Sun and their child ZFS is already suffering from the clutches of the scorned mistress ECC, while the Dell/ Perot and HP/3Com matches are still in their honeymoon period. One partnership though which does look likely to stand the test of time and causes the biggest threat to all others is that of Cisco, EMC and their wonderkid Vmware.

The discussion on how in my opinion these three will monopolise the future of ILM and cloud computing is a blog in itself but my thoughts are best expressed and summarised in a response to a recent conversation I had with a Snr Technical Consultant from an unnamed storage vendor. Scoffing at EMC and what he felt was an erratic spending spree of buying out competitor alongside overhyped marketing jargon which lacked any real focus or gameplan, I begged to differ and pointed out my conclusions:

Point 1: Already established with the DMX, when EMC purchased the Clariion, they acquired arguably the best modular storage array on the market such that still today companies such as Pillar and Huawei Symantec find it hard for their own excellent products to compete with. EMC then purchased Vmware, a server virtualisation platform which is now at the forefront of the technology and continues to innovate and lead the way for other companies to try and follow suit – the recent RedHat Virtualisation platform a prime example.

Point 2: More recently the target based deduplication masters DataDomain have been added to the portfolio to accompany their acquisition of Legato Networker and their source based deduplication tool Avamar. Hence Storage, Server virtualisation and Backup and Recovery are bases well and truly covered.

Point 3: As EMC pushed forward the V-Max alongside flash SSDs, the next approach was towards cloud computing and with cloud computing's major concern being security does it come as no surprise that EMC have also purchased the security firm RSA?

Point 4: By bringing Cisco on board, they have a partner whose clout is already proven in the DataCenter such that based on their IP customer base they were able to muscle themselves in on what was quintessentially Brocade's SAN market. With that in mind does anyone really doubt that the UCS platform won't be purchased on a large scale as DataCenter managers look for the best deal to buy in new IP and SAN switches in these economically testing times?

Hence my final point to the vendor consultant - Erratic these decisions have not been but rather well calculated and planned steps towards DataCenter domination.

What I didn't mention to him is that the EMC marketing machine should not be slurred upon with stains of envy but rather commended. The news that FAST is now available for EMC storage systems has the media jumping in a frenzy despite similar products already being available for some time from other vendors. Back in 2008 I had excitedly completed my training on a great new HDS product called Tiered Storage Manager – but rarely did I see the press mentioning its great value to ILM in the same manner.

So what now of the news that Microsoft and NetApp have refused to join the lonely hearts club and concert their efforts to form a new strategic collaboration to enhance the integration of their virtualised server environments? A three year deal, which will probably extend has been pitched as a natural progression which has no relation to the EMC, Vmware, Cisco partnership , (I stroke my chin dubiously) and one that will focus on better offerings in virtualisation, cloud computing, data management and storage. After losing out on Data Domain to EMC, a partnership with Microsoft is good news for NetApp and a step in the right direction as virtualised infrastructure solutions based on Windows Server 2008 R2, Microsoft Hyper-V Server 2008 R2, Microsoft System Center will seamlessly integrate with NetApp's own systems.

This leads us to Hitachi Data Systems, who often seem to me to be the solitary playboy who refuses to chase but banks on the quality of his product such that he expects 'all' to come to him. No doubt, HDS products are still top of their game in terms of reliability and arguably performance but their current position to have not partnered with anyone runs in the same vein as their lack of drive for marketing. For a company that offer a full range from SMB to enterprise solutions I have often found it very frustrating that their lack of marketing has often led to customers purchasing inferior products or technical solutions from either competing vendors or OEMs – point in case - see how few HDS enterprise arrays are actually virtualised with their own HDS modular arrays behind them.

As for Europe and the Middle East if you have a DataCenter with HDS enterprise storage, chances are they probably have a Hewlett Packard or SUN Microsystems badge leaving most IT Directors and CEOs unaware that they have HDS products. Couple that with the inevitable scenario of the world being 'V-Maxed' there still isn't any news on the long awaited replacement for the USPV. Moreover after recently visiting the Storage Expo in Netherlands, it was almost surreal to see the HDS stand merely offering free sushi and sumo wrestling dolls, while opposite them EMC were giving demonstrations on their DataCenter tool Ionix.

As storage and virtualization begin to nurture more partnerships than a dating agency and resources become aligned to gain more market share, HDS still remain taciturn. HDS already have top quality NAS products which most people still don't know about, top of the range archiving platforms and modular arrays so what are they waiting for? Perhaps there's a great plan on the horizon and they are being strategically patient....I hope so.

It Isn't Over Yet For Fibre Channel Over Ethernet

Advocates of FCoE will immediately mention the benefits of simplified network management, elimination of redundant cabling & switches, reduced power and heat requirements, alongside enhanced data network performance yet despite that still today the FCoE revolution failed to take place. But as the enhancement of Lossless Ethernet begins to really take shape with its promise of a non requirement for TCP/IP overheads, CEE / DCE / EEDC are now looking like viable transports for storage traffic and storage fabrics.

From the moment 10 Gigabit Ethernet became a reality to the mainstream, the emergence of FCoE was a mere formality. Encapsulating Fibre Channel frames over Ethernet networks, FCoE offered the possibility of allowing Fibre Channel to use Ethernet networks while preserving the Fibre Channel protocol. With speeds close to Fibre channel, FcoE now offers companies a tempting cost effective alternative which would support both FC and Ethernet traffic over a single physical infrastructure. Hence storage and network traffic over a single set of cables, switches, and adapters, thus saving in the complexity of managing two physical networks and consequently saving on energy consumption and heat generation.

Furthermore by replacing the FC0 and FC1 layers of the Fibre Channel stack with Ethernet, FCoE allows seamless integration with existing Fibre Channel networks and management software due to the FCoE protocol specification retaining the native Fibre Channel constructs. Hence it comes as no surprise of the news that many vendors are now seriously developing marketing strategies and products which will incorporate the latest and supposedly improved versions of the FCoE standards.

As well as the SAN switch boys Brocade, Cisco and QLogic, vendors such as Emulex, Intel, PMC-Sierra, NetApp and EMC are all looking to develop and market FCoE with new FC and FCOE switches as well as CNAs. Indeed it is the CNAs (Converged Network Adapters) which are the magic behind enabling the connection between the host and FCoE by containing both the functionality of a FC Host Bus Adapter (HBA) and Ethernet NIC on the same adapter.

But FCoE does come along with certain snags as well the obvious one of not being as secure as FC. Firstly load balancing and thus optimal resource utilisation is still an issue due to Ethernet being a Layer 2 protocol and thus leaving FCoE to be unroutable. Hence currently multipathing is still not an approved option. Ironically the problem arises from the advantage FCOE presents with the disbandment of using both Ethernet for TCP/IP networks and Fibre Channel for storage area networks, in favour of one unified network. With Fibre Channel running on Ethernet alongside traditional Internet Protocol (IP) traffic, thus becoming just another network protocol, FCoE operates directly above Ethernet in the network protocol stack, in contrast to iSCSI which runs on top of TCP and IP. Therefore as a result of this FCOE will fail to function across routed IP networks as it is unable to be routed at the IP layer.

Another concern is that once the marketing hyperbole of 10-Gbit is brushed away the truth that remains is that any storage traffic initialised at 10-Gbit will still get dropped onto an 8Gb native FC SAN or 4Gb in the case of most Cisco and Qlogic switches. This becomes more of a point to raise when put in the context that the Fibre Channel Industry Association (FCIA) are showcasing roadmaps for FC which designate that FC will advance from 4->8->16->32 gigabit.

Despite this in an economic climate in which consolidation and cost effectiveness have become keywords, FCOE may be the option that once developed, tested and proven in the mainstream; most customers will be looking to scale out to. With the ability to not only reduce the NICs needed to connect disparate storage and IP networks but also the number of cables and switches as well as the power and cooling costs, FCoE's benefits could well be an option that most companies will now find hard to ignore.

How To Pass The VSphere4 Exam

Having installed VSphere4 on my laptop, cramming the white papers consistently for two months and attending the 'What's New' Course I thankfully passed the VSphere4 exam at the first attempt and hence upgraded my VCP3 status to VCP4. With the relief of passing the exam, I felt it might be useful to pass on some tips and advice to fellow virtual addicts who may be thinking of attaining the certification.

Firstly and most obviously you have to maintain your hands on experience, not just with VSphere4 but also ESX 3.5. I was surprised at actually how little has changed between ESX 3.5 and VSphere4 and it is easy to get caught up with the notion that this exam is only going to focus on the new features of VSphere4 leaving you to possibly neglect reminding yourself of the core skills and concepts you use on your ESX 3.5. While you definitely need to know about the new features such as Data Recovery, the Distributed Vswitch, VApps, Thin provisioning etc. the exam will still pull out some tricky questions related to core stuff such as HA, VMotion and VStorage.

So my tips in preparing for this certification is to firstly give yourself at least 2 - 3 months hands on time with VSphere4 even if you are well versed with ESX 3.5. There are many ways to stick VSphere4 on your laptop (just google ESX on a USB stick and you get some great results) and the Distributed VSwitch is also available on a 60 day trial basis as well as other features.

Secondly I think it's essential to go on one of the courses, although I must admit that the two day 'What's New' course did pack a lot in and could easily be extended to a 3 day session. The course material and labs offer you the chance to link mode datacenters, migrate virtual switches, use DPM on hosts etc something a bit tricky to achieve if you have limited facilities at your work site. I always find you learn more by making mistakes and troubleshooting - something which you can afford to do during the course labs on the course.

Thirdly as well as the course material and configuration manuals, you must download all the white papers and go through them. You will inevitably face questions that refer to information specificaly mentioned in the white papers and no where else, but the good news is that all of these are easily available for download from the VMware site. In addition to this a great book to use and one which will guide you through and also remain as a sound reference is Scott Lowe's Mastering VMware VSphere4 book.

Lastly and what I found to be the most useful of all resources are the blogs and sites of Simon Long and Scott Vessey. There you will find direct links to just about every document you need as well as great tips on how to get the best out of your VSphere4 platform.

In conclusion I preferred the VSphere4 exam to the previous VCP3 exam for several reasons - very few questions if any on silly memory games such as configuration maximums and more questions focused on knowledge of what you actually use and access on a daily basis in your virtual environments. My only gripe is that it's still a multiple choice exam and being a techy freak I long for the day when VMware will go the route of RedHat and offer a fully lab intensive exam - but then again that is probably why they also offer the VXDX!

1.6 Million Reasons To Look Closely At The F5100 Array

It's not often that I get excited over new hardware, well at least not on a daily basis but the new Sun Storage F5100 Flash Array was something I felt compelled to write about.

So what exactly excites me of what is in essence just a JBOD of SSDs? Could it possibly be the delivery of over 1.6 million read and 1.2 million write IOPS? Could it also be that an approximate 100K price tag for 2TB of SSDs works out at as a much cheaper alternative to flash drives which would have ordinarily been in a capsule residing at the back end of an Enterprise array? The answer is "yes" and then some. Not only is it a cheaper option but being based on Single Level Cell (SLC) non-volatile solid-state technology with 64 SAS lanes (16 x 4-wide ports), 4 domains and SAS zoning, the F5100 provides flexible and easy deployment as well as scalability.

In order to really take advantage of the high performance of this array the obvious step is to directly attach it to any read intensive databases such as a datawarehouse. Hence the first step would inevitably be to decipher which data within your read intensive database would be most appropriate to ascertain the maximum performance of the Flash array. To aid you with this SUN also conveniently provide a free Flash Analyzer which provides you those very results by analysing which LUNs are facing the most read IO activity.

So once the pinpointing of the appropriate data has been identified, ZFS can then be utilised to automate the data management and protection. Using the data-integrity features in Solaris ZFS, which checks and corrects block data level checksums, corrupt blocks can easily be identified and repaired. Another option is to look at the major overhead of the ZIL vol of the Zpool of a read intensive database, isolate it to the Flash array disks and see your performance shoot up.

Another nice addition is that the F5100 comes with the user friendly GUI Common Array Manager , (far simpler and quicker to grasp than say HDS' Storage Navigator or EMC's Navisphere. Therefore managing the disks as LUNs, checking their health status and viewing alarms is no different to the normal environment Storage Engineers face on a daily basis.

This is where I see flash drives really taking off and being adopted in the mainstream i.e directly attached SSDs which are a fraction of the price and offer the same if not better performance than there Back End Flash SSD counterparts. A nice product to coincide with the Oracle SUN partnership and an interesting challenge to the Storage vendors who have already sold their soul to the idea that Flash must exist on the Back End...

Live Life To The V-Max

It was early 2009 when I first received word of EMC’s new Enterprise Storage, which at that time was being coined ‘The Tigron’. Rumours were abound that EMC were now going down the same route as the HDS USPV and taking the virtualisation through the array route. Some days later though, the launch of V-Max took place and what was initially thought of as marketing hyperbole did indeed present us with a completely new proposition altogether….

After the slight disappointment of the USPV launch in that personally it didn’t really seem to be a major change from the USP – granted it had 4 Gbps Fibre Channel connections to disk drives and hosts, new processors, support for half-size controller cards etc. I never felt that HDS really stepped up the mark when you consider the fact that the 1152 drive count was the same. So was V-Max also just another beefed up Symmetrix?

Upon observation architecturally V-Max is quite different from previous Symmetrix models even though the same software base is used with the familiar front end/back end functional design. The first significant change is the global cache which is now distributed among the directories’ memories as opposed to the centrally located and separate entity it used to be. Furthermore via the Virtual Matrix technology the directors are now physically interconnected via a dual-star network, enabling director-to-distributed global cache and director-to-director communications.

Another significant change is the underlying Power PC-based processing hardware replacement with Intel x86, coupled with Intel multi-core processors on each director. Thus the functionality of both front-end and back-end directors merge onto one processing complex, enabling local access to hardware for what were only global cache contents.

With up to eight engines which each provide two directors, the V-Max directors each comprise two Intel Xeon 5410 64-bit quad-core processors and their associated main memory and PCIe serial buses. Via the Virtual Matrix each director’s memory is then made available to all the other directors for a maximum of 1 TByte of global memory. And while custom ASICs have been utilized in previous Symmetrix models, by having the Virtual Matrix Interface (the ASIC / System Interface Board) to bridge memory access onto the RapidIO-based Virtual Matrix, the V-Max instead uses a fully standard interconnect between directors and global memory with all the custom hardware encapsulated together.

Furthermore while the predecessor Symmetrix models used two separate networks or buses for data and messages, the new V-Max uses QOS features which prioritize traffic as both messages and data travel over the RapidIO network. The V-Max also boldly claims to give three times the I/O bandwidth of the DMX-4 with double the number of ports.

So is this is a major step up from the Symmetrix? – It certainly seems so and one can only wait with excitement to see what new features EMC will add to this mechanical beauty as they prepare to launch the new V-Max version next year. In the meantime we wait for news on what move HDS will make and whether their imminent replacement of the USPV will be just more of the same or something with more than just a new name.

VM Backups Should Not Be A Virtual Nightmare

As I look forward to attending the upcoming VSphere 4 – What’s New course, I thought I’d share some thoughts on one of the main issues I hope Vsphere will improve upon, namely backups of VMs.

Often people like to stick to what they know, scared to venture into the unknown. Standard backup methods is one of them and it is now leading to the detriment of traditionalists questioning whether they have too many virtual machines on a single server. Scared because the amount of VMs on their single physical server will equate to longer backup times, the traditionalist workaround is to then limit the number of virtual machines placed on an ESX server, henceforth decreasing the overall objective of virtualization. To make matters worse you then find situations where companies have bought extra physical servers to compensate for the backup requirements of the virtual machines – ludicrous and bizarre as it sounds it is happening in a datacenter near you….

From my own limited dealings with colleagues, customers and fellow seminar-geeks which I’ve come across in my travels it’s still surprising to see how few can vouch for their VMware servers being backed up by VCB. Still content with backup software designed for physical servers, I even met one fellow who claimed he’d been to a site where the customer was running their backup software at the physical level inside the ESX server, something which haunted me with nightmare images similar to that of Freddy Kreuger being in charge of cabling Enterprise Storage Systems. The inevitability of having to do daily full backups is a given as any change in a VM results in a modification of the timestamp of its associated VMDK files. Thus there ends up being little difference between incremental or full backups.

Upon attending the ESX 3.5 course and attaining VCP status nearly two years ago, I was desperate to implement VCB and see how great the results would be. Installing a physical Windows server next to the ESX server and thus giving it access to the storage for the VMFS file systems, it was a straight forward process providing access to both block-based (Fibre Channel and iSCSI) and NFS-based data stores. The server would then act as a proxy server to back up the data stores without the I/O having to go through the ESX server. Fabulous I thought, until I realized I was one of the few who was actually calling for its implementation.

Step forward and we see that as technology progresses so do solutions and it was recently that I came across Continuous data protection (CDP) and near-CDP backup products which are being marketed as deduplication-like software. By installing them on the VM the ability to back up virtual machines as in the same vein as a physical server is now possible. But while this constitutes to an advantageous low CPU and I/O load I’ve yet to come across a software that will recover an entire machine, a problem if your VM is deleted.

Like every optimist I do see one promising possibility for future development - the storage systems themselves of which some are claiming to have VMware-aware near-CDP backup built in. Dell EqualLogic, FalconStor Software Inc. and NetApp are already trumpeting this ability. By having VMDKs stored directly on their storage, a tool designed for VMware allows you to tell the system when to run a back up of VMware. Similar to VCB, VMware performs a snapshot which then triggers the storage system to perform its snapshot of the VMware snapshot. By replicating that backup to another box you then end up with your backup, with minimal CPU load on the ESX server and I/O hits on the storage. NetApp has gone one step further with this and uses the VMware tools to do the snapshots and can even dedupe the VMware data including live data.

Surely it’s only a matter of time before the other vendors catch up and do likewise (I give it the usual 6 months.) So as I embark upon my course I look forward to seeing and learning what new strategies are available – I will keep you all posted.