As the customer you have more leverage in this economy than ever before. At the same time - I am sure you are getting pressure from your management to hold off on spending. However, if you have the money - then now is the time to spend it. In a down economy it is the people with cash that get the great deals. Naturally I am not promoting wasteful spending money - rather if you have a need then now is the time for you to get the best arrangement you can. Here are some tips and ideas: 1. Timing does matter. End of month and end of quarter do make a big difference. Everyone knows it and yet we all do the little dance. So do the dance and apply the pressure. They will do the same but remember you have the leverage - it is a buyer's market. 2. Competitive pressure works. Depending on the size of the deal vendors will fight hard to beat their competitors. Again, in this economy pretty much any business is important to them. Having two or three major competitors fighting it out for your business can drive down cost significantly but its hard work. Don't let yourself give up because you are tired. Keeping grinding those wheels - it can make a big difference. It is also important for you to realize that not all competition is equal. On some level every competitor is a threat but there are certain competitors that have near-religious wars with one another. If you can identify the competitors that are arch-enemies then you can leverage that in your favor. 3. Get more for your dollar. Perhaps the overall cost isn't as important to you as other things - e.g. getting more software, services, training and support can be negotiated as a part of the arrangement. 4. Always negotiate additional capacity at the original point of negotiation and any software licenses that may be bounded to additional capacity. Don't wait until you need the capacity - if you do then you lose leverage. Now capacity goes down over time so you have to make sure that you base pricing on a percentage discount off of list pricing. But the point is to get the best discount now and make sure it applies to all future acquisitions. 5. I suggest you read an end user blogger, Storagebod wrote a blog on negotiating storage that has some good insights as well. I suggest you check it out - Deal or No Deal? 6. I think that negotiations should be done with a strong position but not to the point of being combative. I suggest speaking to the vendors with good intentions - "We really want to work with you but we are doing our homework." You also don't want to make the vendor feel like they got a bad deal - because often times it is the sustainable partnership that makes the difference between bad and good and also between good and great outcomes. If you negotiate smartly and firmly then you should wind up getting respect versus being seen as a jack-ass. 7. A couple of other things to consider is being a reference for the vendor. There are two kinds - one that will only talk privately to other potential customers or you can be a public reference as well. You many not be able to get a bigger discount but you will certainly get a lot of attention when a problem occurs in your data center with their products. "Hey, I am a public reference for you guys and I have an interview with the press in an hour to discuss your product but funny it isn't working. Can you help me out with that?" Chances are you won't ever have to vocally use this card but its power is palpable. 8. In my first blog in the Negotiating Storage Part One - I discussed the idea that you should think about other stuff that you need in addition to storage. This is where your channel partner comes in - get them to bundle everything up for you for discounts across the board. But don't forget that all this stuff has to be put together. And often, there are nuances with the storage product with this other stuff that will create walls and delays. Make sure that they storage vendor understands what you are doing and negotiate ahead of time you need their support and experience to help with that other stuff.
The good news / bad news is there are a lot of storage systems to choose from. In terms of cost - it is important not just to consider the cost of the actual system but power and cooling remain a big issue and of course ease of use needs to be considered. Perhaps what is most important in this day and age is how optimized is the storage system? Here are some storage technologies that should reduce the overall cost of your system: 1. Thin provisioning. Now just about every vendor says they support thin pro but some vendors are better at it than others. Do your homework if this is an important feature to you. Thin provisioning can significantly reduce the amount of allocated but unused capacity. Since allocated but unused capacity is typically between 50-60% using traditional provisioning methods - it is pretty crazy for you not to consider thin provisioning as a top priority feature. And look - it was first on my list. 2. Logical snapshots. Everyone has snapshots and so this should also be one of those no-brainer technologies. Again, some vendors are better than others but snapshots should be a requisite. These are logical and space-efficient copies of primary data. 3. Writeable snapshots, Not everyone has writeable snapshots. With read-only snapshots you have to mount the entire snapshot volume and consume the same amount of capacity as the original volume that you were snapping. Writeable snapshots allow you to create an active read/write snapshot directory that only consumes new and different data. Much more efficient and cost effective than mounting a read-only snapshot volume. 4. Growing and shrinking of volumes. This is different than thin provisioning. This requires integration with the host O/S and or file system. The idea is that you can grow or shrink a volume as needed. Not every storage system supports this capability but it can be extremely useful. I especially like the "shrink a volume" capability - reclaiming space is a big part of the battle. Thin provisioning does not solve the shrinkage issue (am I allowed to say that?). 5. Compression or data deduplication. Not many primary storage systems support any level of compression or data deduplication - but even if you just get a 2 to 1 capacity optimization - it is worth looking into. In this day and age - not having this capability seems pretty silly. But I can't count on one finger the number of storage system vendors that provide this for primary storage (NetApp). 6. Tiered storage. Most storage systems do support different drive types including FC and SATA or SAS and SATA. Again, a no-brainer that you should consider. Some storage systems only support SATA and if they can meet your reliability and performance needs that is a good thing - just make them prove it. 7. Intelligent tiered storage. Having different types of disk drives is one thing but being able to move data between these tiers is another. I define intelligent tiered storage as the ability to move data transparently between different drive types and RAID levels. You should also be able to create copies of primary volumes on different drive types and RAID levels. Most storage systems that support intelligent tiering do it at the volume level. I do have to call out Compellent - since they can do it uniquely at a block level using something they call Data Progression. 8. Capacity flexibility. Some storage systems are limited in the amount of capacity they can support on a controller basis. They hit a ceiling pretty quickly. For the majority of you having a 10 TB or even 60 TB storage system might be fine. However, if you want 200 TB or more in a single storage system then the list goes down. Sure only 10% or 20% of your primary data may be active at any given time but you may want to keep that other 80%-90% managed by the same storage system on lower cost disks without having to buy more controllers. Buying more storage controllers to get more capacity can be a costly proposition. And be careful about the marketing number versus reality - capacity support is overhyped. How can you find out? Make them give you customer references that validate they support the amount of capacity you are looking for. 9. Disk drive spin down. This can also be a part of an intelligent tiered storage system. Spinning down drives can reduce power and cooling costs - and should be able to extend the life of your disk drives. Again, a very small percentage of your data is active - so why not spin down the drives that store inactive data? 10. SSD technology. Wow, who would ever thought that the dud of the storage world would start to look sexy? You have to thank EMC for this along with their partner STEC. The concept is that there are times that IT professionals need to support a large number of disk drives in the storage system and wide strip across them in order to get higher performance - even if they don't need the additional capacity. At that point you are paying for spindles and not storage. This can get costly with the direct cost of the disk drives and the consumption of power, cooling and floor space. However, using SSD in a disk drive form factor you can significantly reduce the number of disk drives to obtain the same performance. However, SSDs have been around forever so why are they getting so much attention now? Historically SSDs haven't been good at all types of I/O but new caching technologies make them better at mixed workloads. Additionally, with the size of disk drive capacities escalating - buying all that capacity just for performance is hard to justify. And of course power and cooling is such a big issue that the higher cost of SSDs can be offset with a reasonable ROI. 11. Unified storage. Most storage systems support FC and iSCSI. Some support iSCSI and NAS. Fewer support FC, iSCSI and NAS. And now FCoE is being added to the mix. I like unified storage because it gives IT professionals more flexibility and it can lead to further consolidation of resources. 12. External storage virtualization. This is the ability for the storage controller to manage external storage systems as well as its own internal capacity. The HDS USP family does this (I blog for them as well). Falconstor, IBM and NetApp support appliances that perform this function - except that the IBM SVC does not support is own internal capacity but all external. External storage virtualization does a number of things for you including intelligent tiered storage, heterogeneous data replication, transparent data migrations, and extends the life of other storage systems you may have on the data center floor. I am interested in any other features, capabilities and ideas that you might have for great efficiency of storage infrastructure.
Solving the Enterprise Search Dilemma. Take a few minutes to read it. Search must and eventually will become requisite in the data center - but we just haven't gotten there yet because of a gap in the core competencies needed.The ability to provide Enterprise-class search and indexing requires two very different core competencies. The first requirement is to build a platform – IT infrastructure – to address the needs of the Enterprise. These include massive amounts of content that is stored on heterogeneous storage that is most likely geographically dispersed. How do you index all of the existing content – which consists of hundreds of terabytes and perhaps even petabytes – while new data is created continuously? How long will it take the solution to catch up? Days? Weeks? Months? Years? Ever? In my latest Digital Reef blog - Unstructured Data - I examine what is needed for a True Enterprise Search solution -
Many of you have a great deal of experience negotiating network storage. However, there are a number of you that are just learning the ropes. I am writing a blog series on this - the first one is figuring what you really need. Please provide comments and feedback. The first thing is to really determine what you need. You would be surprised how often people tend to over buy on storage. This includes the type of storage systems, the amount of features and the amount of capacity. This may come as no surprise during a "rich" economy but we obviously do not have this luxury today. I have worked with IT professionals that have bought Enterprise-class storage systems versus midrange when they really didn't need to. I have seen people implement FC networks when they could have used iSCSI. I have worked with folks that used FC drives when they could have used SATA or a combination of FC and SATA drives. Let's take the first one. I have seen companies spend way more than they had to because they invested in an Enterprise-class storage system when they could have implemented a much lower cost midrange solution. Now, there may an application or two that are so important to your business that they extra cost - the "insurance" - is worth the price but that doesn't mean that you have to put all of your applications on that tier. However - I would strongly argue that many midrange storage systems can meet the needs of most applications from high availability, feature and performance perspective these days. If you are working for a large company - you pretty much have figured these things out. If you are a smaller company putting a new storage network - then consider midrange as your core storage solution. You have lots of choices out there. There should be no question any longer that iSCSI is a viable alternative to FC for storage networking. One could argue that is better on many levels. There is both capital and operational cost savings to be had by implementing iSCSI. I question any small to midsize company implementing FC storage networks at this stage of the game. Yes, the larger companies will continue to but they have legacy to deal with. There are storage systems that have SATA only in them dealing with mission-critical applications and they are working fine from both a performance and reliability perspective. Like iSCSI, SATA drives have been proven over the years. But again, it doesn't need to be an all or nothing proposition - if you feel you need some FC drives for a specific application or workload - most storage systems allow you to combine FC and SATA drives. Again, sounds obvious but you would be surprised that many people still default to FC drives. How much capacity do you really need? We have seen environments that have only 20% capacity utilization and others that have 80%. This topic requires a blog of its own - so I will be following up on this. Here are some things to keep in mind - ask your vendors how they can optimize capacity utilization - what tools do they have? Also - as part of negotiation tactics - make sure you negotiate all future capacity acquisitions at this stage of the game and not later when you have little to no leverage. Here is the deal - something perhaps not as obvious to consider - a storage system is not standalone - it lives within an ecosystem that includes servers, applications, data and system protection, and networks. As a small or midsize company - you want to leverage the system integrator to take care of the overall solution holistically. Are you getting new servers as you acquire storage? Have you considered how you are going to backup the new storage system? What user applications will you be using? Are you going to implement server virtualization? Are you considering a disaster recovery strategy using remote replication? Many IT professionals like working directly with vendors but most of them can't help you with what you need end-to-end. Therefore it is important to figure out what you need end-to-end and work with someone that can help you from this perspective. Another important thing to consider is the software features that the storage system provides. Features including snapshots, wide stripping and thin provisioning are valuable and important. I would go as far as saying that these three features are becoming requisite. Additionally - ease of management is critical and in this day and age having an overly complex storage system should not be tolerated. The next blog entry will discuss important and valuable features within a storage system - many of which will directly impact your costs.