Originally Posted Dec 14, 2012<\/p>\n
This week I was very privileged to go to Salt Lake City to\u00a0Fusion<\/strong> IO<\/strong> headquarters and get a deep dive on their technology and how it differentiates from other competitors in the high speed, low latency storage market.\u00a0\u00a0(Which is really starting to just be the general storage market these days instead of something niche.)\u00a0\u00a0It was super neat to go there with a bunch of my buddies from Cisco and I can\u2019t thank them enough for having us and treating us so go.\u00a0\u00a0This meeting was brought about primarily because\u00a0Fusion<\/strong> IO<\/strong> and Cisco have introduced a new Mezzanine\u00a0Fusion<\/strong> IO<\/strong> card for the B series blades.\u00a0\u00a0Specifically:\u00a0\u00a0The B200 M3 and B420 M3.\u00a0\u00a0The card is the same as the ioDrive2\u00a0\u00a0but in Cisco blade server form factor.\u00a0\u00a0This first version is 758GB drive.\u00a0\u00a0We had a great time and and learned a ton.<\/p>\n Fusion<\/strong> IO<\/strong>\u2019s main product is branded as ioMemory.\u00a0\u00a0Its marketed as a new memory tier.\u00a0\u00a0The co-founder, David Flynn had the idea of taking Flash Memory,\u00a0\u00a0putting it on a PCI card, slap a basic controller on it and putting it into the server.\u00a0\u00a0The server would then see this as a hard drive.\u00a0\u00a0By not using legacy protocols of SAS or SATA and using their own protocol with software, they were able to get\u00a0IO<\/strong> latency down to microseconds from milliseconds.\u00a0\u00a0Couple this with flash memory and it translates to more IOPS<\/a>, which means applications that normally have to wait for disk reads and writes can do it on orders of magnitudes faster.\u00a0\u00a0One of the examples they cited said that customers were getting 20 times the performance with these drives compared to using standard disk drive arrays.\u00a0\u00a0(Not 20% better, but 20 x)\u00a0\u00a0From that above linked Wikipedia article<\/a> it shows that fastest 15k RPM hard drive will only get around 200 IOPS.\u00a0\u00a0Compare that to a\u00a0Fusion<\/strong> IO<\/strong> card that same article shows 140,000 IOPS.\u00a0\u00a0(My notes also say that they are getting 500,000 IOPS, I\u2019m not sure which is correct but the idea is that its blazing fast.)<\/p>\n If you aren\u2019t familiar with the state of the data center, and I commented on this in myVMworld 2012 post<\/a>, storage is one of the biggest problem.\u00a0\u00a0Numerous blogs, articles, and talks show that storage is the biggest bottle neck and the largest expense in the data center.\u00a0\u00a0Duncan Epping commented at VMworld on the topic of performance that \u201cThe network is usually blamed but storage is usually the problem\u201d.\u00a0\u00a0 There is a famine happening in our data centers.\u00a0\u00a0Applications are starved for data.\u00a0\u00a0They are waiting and waiting to get their data from disks.\u00a0\u00a0Applications today are like a bunch of hungry children at home whose moms went to the store to get food and are all stuck in traffic with other moms taking a long time to make that round trip.\u00a0\u00a0Storage\u00a0IO<\/strong> performance has not kept up with the spectacular rate that processing power has improved over the last decade or so.<\/p>\n What we have been doing for the last 10 years (me personally and others) is designing storage systems that we think will meet the performance requirements.\u00a0\u00a0When we get up and running we soon find that the storage system doesn\u2019t meet the performance needs so we throw more disks at them until it does.\u00a0\u00a0Soon we have tons more capacity than we need and a bigger footprint.\u00a0\u00a0I\u2019m not alone in this.\u00a0\u00a0This is standard practice.\u00a0\u00a0Commenting on this, Jim Dawson, the Vice President of world wide sales wrote for all of us to see:\u00a0\u00a0DFP = RIP.\u00a0\u00a0This he said means:\u00a0\u00a0Disks for performance is dead.\u00a0\u00a0He also mentioned that his customers when he was at 3PAR were adding so many disks for performance that they asked him to make smaller disk sizes because they didn\u2019t need capacity, they needed performance.<\/p>\n Flash Memory to the rescue<\/strong><\/p>\n The reason you are probably hearing so much about flash memory now and not before is because the price of flash memory has fallen below the price of DRAM (the kind of memory that when you pull the power power it, it forgets everything that was in it).\u00a0\u00a0Flash memory, specifically NAND flash, is the flash that\u2019s used in\u00a0Fusion<\/strong> IO<\/strong>, SSDs,\u00a0\u00a0SSD arrays, and pretty much everything you see out there that\u2019s called flash storage.\u00a0\u00a0 This type of memory when you pull the power doesn\u2019t forget which bits were flipped to ones or zeros.\u00a0\u00a0NAND flash are the building blocks for nearly all the fast storage you\u2019ve been hearing about.\u00a0\u00a0From\u00a0\u00a0people making USB thumb drives, SSDs, PCI SAS, Violin, or Texas Memory Systems (now IBM) and make arrays with them using their own controllers, they\u2019re all using NAND flash.<\/p>\n The difference is how the flash is accessed.\u00a0\u00a0SSDs go through the SAS or SATA controllers that add significant over head.\u00a0\u00a0That makes it slower since those are legacy protocols used for hard drive technology.\u00a0\u00a0But if you have one in your mac book pro like I hope to have soon, then you are not complaining and its just fine.\u00a0\u00a0Most of the Flash storage solutions out there are based on using SAS\/SATA protocols to access flash storage: Nimble storage, whiptail, etc. Its more simple to develop because the protocols are already defined and they can concentrate on value add at the top, like putting more protocols or better management tools in it.<\/p>\n Fusion<\/strong> IO<\/strong> has two advantages over these technologies.\u00a0\u00a0First, since they are on the PCI bus, they are closer to the processor so its much faster.\u00a0\u00a0Second, they don\u2019t have the overhead of a controller translating older protocols.\u00a0\u00a0There\u2019s a driver that sits on the OS that manages it all.\u00a0\u00a0Since they don\u2019t go through the standard protocols they can also add better monitoring tools and even add more on top of that to innovate cool solutions.\u00a0\u00a0(ioTurbine is an example of this that I\u2019ll get to in a minute)<\/p>\n Fusion<\/strong> IO<\/strong> secret sauce<\/strong><\/p>\n The ioDrive2 card is main product.\u00a0\u00a0Its a PCIe card with a bunch of 25nm NAND flash chips on it.\u00a0\u00a0We had this amazing\u00a0Fusion<\/strong> IO<\/strong> engineer named Bob Wood come in and talk to us about how it works.\u00a0\u00a0He schooled us so hard I thought I was back incollege<\/a>.\u00a0\u00a0We were worried we were going to get more marketing but in the words of @ciscoservergeek: Our expectations were amazingly surpassed.<\/p>\n Flash memory has what\u2019s called an Erase Block.\u00a0\u00a0This is the smallest atomic unit that can be written.\u00a0\u00a0As flash gets smaller having 3 or more electrons leave, or somehow get disturbed will cause the erase block to flip a bit and be wrong.\u00a0\u00a0The controller software is then always looking to make sure things are still the way they should be.<\/p>\n