|
Solution Search:
BUYING GUIDE
The following cross-section of midrange NAS arrays was selected based on input from industry analysts and...
More...
STORAGE ARTICLES
Promise Technology Inc. extended its VTrak storage system product line with a new appliance that includes storage virtualization, snapshots, mirroring, data replication, failover and thin provisioning through partners.
The VTrak S3000 is a 1U system that supports 8 Gbps Fibre Channel as well as Gigabit and 10 Gigabit iSCSI, and scales to scales to 32 TB in one node. The system is also available in multi-node configurations. Pricing for the VTrak S3000 starts at $36,000. The new features come from partners Atempo, DataCore Software, FalconStor Software, NEC, Sanrad, Tarmin Technologies and Vicom Systems... More... EMC Corp. picked up another piece of the IT infrastructure stack outside of storage today when it acquired privately held data warehousing vendor Greenplum Inc.
Greenplum's product platform includes Greenplum Database, a massively parallel processing (MPP) database for analytical processing; free single-node version of Greenplum Database, and Greenplum Chorus, which is aimed for enterprise clouds and includes self-service provisioning, data collaboration and data services. Greenplum's products run on x86 hardware. EMC did not disclose the price for the cash transaction, which the vendor said... More... EMC Corp.'s decision to sell its Atmos cloud storage service solely through service providers instead of setting up its own service comes as good news for its service provider partners. The timing couldn't be better for one partner, Hosted Solutions, which launched its Atmos-based backup service last week just days before EMC made its decision public on its website.
Hosted Solutions' Stratus Cloud Storage is one of three Atmos-based commercial services so far, along with AT&T's Synaptic Storage as a Service and Peer 1 Hosting's CloudOne Storage. Hosted Solutions has Atmos frames in two of its five data... More... Data protection service provider Data Storage Corp. (DSC) has acquired backup and recovery provider SafeData LLC, allowing DSC to expand its products to include backup, disaster recovery, digital archiving, virtual hosting, and e-discovery.
SafeData will operate as a division of Data Storage Corp. and will continue to serve its existing clients with SafeData/HA, SafeData/DR and SafeData/TR products. Peter Briggs will continue as president of the SafeData unit and will hold the position of executive vice president at Data Storage. Dot Hill launches new 6 Gbps SAS storage arrays |
STORAGE BEST PRACTICES
These are still early days for cloud storage, but we've already gleaned valuable best practices from administrators and other experts for getting the most from a move to the cloud, whether you're looking to do it today or down the road:
Best practice #1: Scrutinize service-level agreements Proceed with caution when it comes to getting a service-level agreement (SLA) from a cloud provider. That means read the SLA closely before committing. "There are a few major providers offering SLAs that are very vague about things like guaranteed recovery and assured destruction of data," Beth Israel Deaconess Medical Center (BIDMC) storage architect Michael Passe said at a Storage Decisions session last year. "You want to look behind the wizard's curtain to see what is really there." Lauren Whitehouse, a senior analyst at Milford, Mass.-based Enterprise Strategy Group (ESG), said data access is one area that bears close examination in an SLA. "Generally, SLAs have to do with access to the service, not to data," she said. "Generally, the service has to be down more than 10 minutes before it's considered an outage, so two nine-minute outages in an hour don't count as an outage. If there's an outage of the service, they just adjust the bill -- that's the kind of game that gets played. You have to ask, 'What about access to data?'" Best practice #2: Follow your business needs Lantmännen, a collective owned by 40,000 Swedish farmers, saved more than $6 million in the first year after building an internal private cloud with EMC Corp. storage and Riverbed Technology WAFS devices, said Dennis Jansson, Lantmännen's chief security officer. Jansson said users choose what type of application they need through a web interface, and each service has a fee, SLA and integrated enterprise security management application. "We're able to actually follow business needs," Jansson said of the cloud. "It doesn't make decisions on applications the users need." He called the cloud "an easier way to say consolidation, virtualization and standardization." Best practice #3: Repurpose your own resources Online advertising sales rep firm Gorilla Nation Media LLC built an external customer-facing cloud and an internal cloud for employees by using servers it already owned along with cloud vendor ParaScale Inc.'s Hyper-scale Storage Cloud software to build an object-based clustered NAS system for unstructured data. Alex Godelman, vice president of technology at Gorilla Nation, said the cloud replaced a more expensive NAS setup. "To grow the internal cloud, we just add more nodes," he said. "The design of the system is also very simple -- we just kind of use it. And it allows us to breathe some life into a huge existing investment, which means we created the system virtually for free." Best practice #4: Prepare for the future Even if you're not ready for the cloud now -- or the cloud's not ready for you -- start thinking about how it may help you down the road. Charles Shepard, director of systems architecture at the MGM Mirage in Las Vegas, said he will consider an external private cloud when technology advances make it feasible. "When Fibre Channel over Ethernet [FCoE] becomes completely adaptable and adopted over the next five years, and when it is completely standardized, that is the pathway to develop a full cloud outside our data center," he said. "If you have a big enough pipe, like 10 Gigabit Ethernet [10 GbE] or even 100 [Gigabit] Ethernet, you might be able to take a database and write from it to the cloud." He said FCoE would be well suited to multitenancy, which is a crucial component of the cloud. "It inherently subsegments networks for internal and external multitenant environments," he said. Best practice #5: Beware of hidden costs Cloud storage providers will tell you the basic cost per gigabyte of cloud storage up front to help you figure out how much it will cost you per month depending on the amount of data you need to store. But these basic costs are only part of the picture, and providers may also charge extra for data transfers, metadata functions, or copying and deleting files. And don't forget the costs of connecting to the cloud, perhaps with a T1 line.
For more on cloud storage: 1. Find out why the evolving cloud storage market has users weighing their enterprise data storage options 2. Discover why external cloud storage appeals to smaller firms, but large enterprises remain cautious 3. We explain how internal private cloud storage makes its way into larger enterprises 4. Read why not everyone thinks the future is bright for clouds Learn best practices for getting the most from cloud storage, including what to look for in a service-level agreement (SLA), using existing resources to set up a cloud and how to avoid hidden pricing. The iSCSI storage-area network (iSCSI SAN) has been discussed and debated for much of the last decade, but iSCSI has finally come into its own as a networked storage underpinning for virtual server environments, analysts say.
Enterprise data storage vendors such as Hewlett-Packard (HP) Co.'s LeftHand Networks and Dell EqualLogic have cited the additional costs of networked storage requirements as a barrier to server virtualization adoption for some customers in positioning iSCSI SAN products for that purpose, and tout the relatively low cost of iSCSI SANs compared with Fibre Channel (FC). But according to Jeff Boles, senior analyst and director, validation services at Hopkinton, Mass.-based Taneja Group, there are some technical considerations that make iSCSI more appealing for virtual servers as well. "A lot of engineering went into Fibre Channel based on the assumption of one host per port," Boles said. "iSCSI has virtualized access anyway, over an IP connection, and has had more engineering around multiple-host contention and various queuing patterns." While Ethernet networks and the basic best practices for iSCSI SANs are generally well understood by now, if you're looking to deploy an iSCSI SAN to support server virtualization, experts say there are some different factors to keep in mind than when connecting physical servers via iSCSI. Here are five best practices for using iSCSI in a virtual server environment. Best practice #1: Look beyond basic iSCSI In the years since iSCSI first came on the scene, products have had time to mature and develop, adding specialized features along the way. In the meantime, iSCSI-related products have proliferated to the point where software-based iSCSI initiators and targets can be had completely free of charge. iSCSI SANs can be built using commodity server hardware and open source software as well. But Boles said iSCSI specialists, like HP's LeftHand or Dell EqualLogic, are charging a premium for advanced features such as integrated VMware snapshots. Other iSCSI SAN vendors, such as EMC Corp. and NetApp Inc., offer unified storage arrays with various options for connecting servers, including iSCSI. Disk arrays from storage specialist vendors also often have features like quality of service and virtual machine-aware management consoles. The iSCSI network these arrays are attached to can also make a difference, Boles said. "If you have the right infrastructural underpinnings, for example a well-built, fully managed Cisco environment, you can apply more sophisticated and granular policies to virtual servers." On the other hand, some of the most advanced iSCSI deployment methods aren't really necessary for a virtual server environment where cost and consolidation are primary factors in purchasing decisions, countered Greg Schulz, founder and analyst at Stillwater, Minn.-based StorageIO Group. As data grows and 10 Gigabit Ethernet (10 GbE) looms on the horizon, some industry experts see technologies like TCP/IP offload engines (TOE cards) coming into play. But users should balance the availability of these performance enhancers with their original rationale for deployment, Schulz said. "If low cost is the reason I'm deploying iSCSI, I'm probably not going to invest in hardware adapters. Instead, I might want to enable jumbo frames and quality-of-service features through software." Best practice #2: Consider where iSCSI targets should live in the virtual environment on an application-by-application basis For VMware environments specifically, "It used to be users had to make a tough choice," Schulz said, between VMware's clustered file system (VMware vStorage VMFS) or raw device mapping (RDM). Before Version 3.5, VMFS offered features like VMotion, but RDM was sometimes the only way to continue to use value-added features of storage arrays like snapshots and virtual provisioning. While this is no longer the case today, Brian Garrett, vice president of ESG Labs at Milford, Mass.-based Enterprise Strategy Group (ESG), said users should still evaluate where to place the iSCSI target in the infrastructure for performance and manageability reasons. They have a choice of deploying the target as either a virtual disk at the hypervisor level, allowing the server virtualization software to handle calls to the back-end storage through a virtual hard disk layer; or at the disk array, providing somewhat speedier block-based access to the back-end storage. "The decision will depend in part on what you're already used to," Garrett said. "But block-based apps like SQL databases, for example, work well with raw disks, and would probably be suited to the pass-through or raw mode." Best practice #3: Rethink network and cabling designs "One thing users often don't think about is the way iSCSI can give you freedom from past paradigms," Taneja Group's Boles said. Storage pros are used to the Fibre Channel world, where a monolithic disk array is attached via a complex series of switches and cables to servers in a separate aisle of the data center. With an increase in scale-out and commodity-hardware-based iSCSI SAN architectures, Boles said a new networked storage deployment might also be a good opportunity to rethink the data center layout. "With some of these iSCSI systems, you can interleave the storage with the server farm, and get the storage closer to the server environment without as many long cables." Rethinking the physical placement of resources in the data center can help resolve issues with overloading parts of the network. "You don't have to shove I/O down a big trunk and then fan-out to the entire infrastructure – interleaving can avoid these bottlenecks," he added. Best practice #4: Be mindful of monitoring Boles and Garrett both emphasized that the new virtual world requires new virtualization-aware monitoring tools throughout the data center infrastructure, particularly as highly portable virtual machines (VMs) move around the network. "When you get into a virtual environment, performance monitoring and tuning become a lot more important," ESG Labs' Garrett said. "In the physical world it was easier to make sure you had the right number of actuators to avoid overconsolidating and violating basic storage guidelines." Added Taneja Group's Boles: "It's easier to implement monitoring from Day 1 than to go back and retrofit a network fabric with monitoring tools; make purchasing decisions with this in mind." Best practice #5: 10 Gigabit Ethernet remains a ways off The next boost in Ethernet bandwidth will probably improve iSCSI performance and offer more network consolidation opportunities within data centers, and the transition to 10 Gigabit Ethernet will begin imminently, according to Rick Villars, vice president, storage systems and executive strategies at IDC in Framingham, Mass. "This will be the year server vendors tell people to go to 10 Gigabit Ethernet," he said. But Villars urged caution when it comes to porting iSCSI SANs to 10 GbE networks too soon, particularly if you're dealing with implementing a virtual server environment already. "You have to decide whether iSCSI is the first or the last thing you want to bring on [to a new 10 GbE network]," Villars said. "Since it's in the early stages, I wouldn't want to go out and start with an iSCSI SAN on [10 GbE] yet." Storage experts offer best practices on how to maximize iSCSI SAN performance and efficiency in virtual server environments, where the protocol has found its biggest audience. No matter how much storage capacity you squeeze into an array, it's just a matter of time until that space is completely filled. Users everywhere are challenging existing storage resources with applications that proliferate media-hungry data files. This is where SAN expansion technology comes in.
This handbook takes an in-depth look at SAN expansion technology and offers advice on the many elements involved, such as integrating SAN and NAS, switch upgrades, disk array replacement and capacity planning tools. Best Practice No. 1: Expanding or replacing disk arraysEven the biggest disk array eventually runs out of storage space. Tiered storage? Data deletion? No matter. At some point, you'll still need to expand or replace your SAN. Here are eight best practices for disk array expansion or replacement. Best Practice No. 2: Switch upgrades and replacements Best Practice No. 3: Host bus adapter upgrades and replacements Best Practice No. 4: Selecting a storage capacity planning tool Best Practice No. 5: Integrating SAN and NAS Best Practice No. 6: Integrating iSCSI and Fibre Channel, and make it work Best practices for SAN expansion include options for integrating SAN and NAS or iSCSI and Fibre Channel, evaluating capacity planning tools, as well as upgrading switches, disk arrays and HBAs. SECURITY TIPS
What you'll learn: Find out how Atrato, DataDirect Networks, Panasas and Xiotech use RAID technology in self-healing systems and also deal with hard disk drive (HDD) failures.
Self-healing systems from a handful of data storage vendors come with the promise of little or no maintenance, and seek to proactively heal hard disk drives (HDDs) after a hard disk drive failure. But depending on which self-healing system you examine, you'll find different approaches to data migration and recovery, as well as disk repair. In this expert tip, Mark Staimer highlights these various approaches to using RAID technology in self-healing storage systems. Atrato V1000 Atrato Inc.'s Velocity1000 (V1000) uses a self-healing technology called Fault Detection, Isolation Recovery (FDIR) in combination with Atrato's Virtualization Engine (AVE). FDIR watches component and system health, and adds self-diagnostics and autonomic self-healing, but it doesn't attempt to remanufacture or recondition HDDs in place as Xiotech does. Atrato puts 160 2.5-inch SATA drives in a 3U system called SAID (self-maintaining array of independent disks). Atrato uses its extensive SATA drive performance database of operational reliability testing (ORT) to monitor the installed drives actual performance to detect SATA HDD deviations. Atrato also deals with HDD faults by first attempting to repair the faulting HDD sectors (although not with manufacturer-level reconditioning, remanufacturing or component recalibration). If the fault or non-recoverable read error can't be repaired, the sector is isolated and only the affected data is reconstructed and remapped to virtual spare capacity. If a disk drive completely fails, it's reconstructed and remapped to the virtual spare capacity. Atrato reduces the number of rebuilds and rebuild times by reconstructing only affected data on virtual drives. Atrato backs its technology with a three-year warranty. DataDirect Networks' S2A DataDirect Networks Inc.'s S2A technology heal-in-place approach to disk failure attempts several levels of HDD recovery before a hard disk drive is removed from service. It begins keeping a journal of all writes to each HDD showing behavior aberrations and then attempts recovery operations. When recovery operations succeed, only a small portion of the HDD requires rebuilding using the journaled information so rebuild times are reduced and a service call may be avoided. Panasas ActiveScan Panasas Inc.'s ActiveScan technology continuously monitors HDDs and their contents to detect problems. ActiveScan monitors data objects, RAID parity, disk media and the disk drive attributes. When a potential problem is detected, data is moved to spare blocks on the same disk. Future HDD failure is predicted through the use of HDD self-monitoring analysis and reporting technology (Smart) attribute statistical analysis, permitting action to be taken to protect data before a failure occurs. When an HDD failure is predicted, user-set policies facilitate pre-emptively migrating the data to other HDDs, which eliminates or mitigates the need for reconstruction. Xiotech ISE Xiotech Corp.'s Intelligent Storage Elements (ISE) is a good example of self-healing storage. ISE tightly integrates RAID and HDDs, and combines them into a single storage element. Xiotech engineered ISE to resolve most RAID rebuild issues by eliminating 67% to 90% of the rebuilds. It starts by reducing HDD faults by proactively healing hard disk drives before a fault occurs using similar HDD reconditioning algorithms employed by the factory. It also uses advanced vibration controls and sealed systems called DataPacs to reduce outside influences from causing HDD faults. When a fault does occur, it reacts by providing remedial component repair within the sealed DataPac using methods similar to what the original manufacturer uses. It analyzes power cycles, recalibrates components, remanufactures the HDD, and migrates data when required to other sectors or HDDs. If the fault persists, ISE will isolate just the non-recoverable sectors and then initiate data reconstruction only for the faulted HDD sectors. So there are far fewer rebuilds and, when one is required, there's much less to reconstruct. In addition, it's all automated so no manual intervention to pull failed drives is required. The result is equivalent to a factory-remanufactured HDD with only the components that are beyond repair taken out of service. The downside to this transformational technology is that it has higher up-front costs, although it lowers the total cost of ownership (Xiotech provides a five-year warranty). Other vendor approaches LSI Corp. and NEC Corp. both detect HDD sector errors while allowing operations to continue with the other drives in the RAID group. If an alternative sector can be assigned, the HDD is allowed to return to operation, avoiding a complete rebuild. Performance is maintained throughout the detection and repair process. This is a limited self-healing technology that reduces the number of rebuilds and helps maintain performance. 3PAR's InSpire Architecture is engineered to sustain high performance levels by leveraging advanced HDD error isolation to reduce the amount of data that requires reconstruction, and taking advantage of its massive parallelism to provide rapid rebuilds (typically fewer than 30 minutes). The system uses "chunklets" in its many-to-many drive relationships. That same massive parallelism allows 3PAR to isolate RAID sets across a multiple drive chassis to minimize the risk of data loss if a chassis is lost. BIO: Marc Staimer is a frequent contributor to TechTarget. This article originally appeared in Storage magazine. Find out how Atrato, DataDirect Networks, Panasas and Xiotech use RAID technology in self-healing systems and also deal with hard disk drive (HDD) failures. What you'll learn: Storage resource management tools can help you manage all of your physical and logical storage resources, devices, disks and files. Here's what you should know before you invest in SRM tools.
Storage provisioning -- tracking how much data storage you have available and then trying to apportion it to various users and groups -- is a fundamental activity for getting the most out of a storage-area network (SAN). SRM tools can play an important role in accomplishing that goal, but choosing the right one can be tough. For starters, the definition of storage resource management (SRM) is too much of a catch-all to be much help. Depending on where you find your definition, you'll see SRM is essentially the management of all physical and logical storage resources, devices, disks and files. But for data storage managers, SRM tools offer the important ability to have detailed capacity reporting. That includes change management and chargeback. This focus is crucial today, said Enterprise Strategy Group senior analyst Bob Laliberte, as data storage environments become increasingly more dynamic and "cloud-like." The biggest recent gain in the sphere of SRM tools has been the ability to optimize even beyond thin provisioning, Laliberte said. Thin provisioning allows different users to be allocated the same physical storage on the assumption they won't all need it at the same time. "Like a bank, if everyone comes in and cashes their checks all at the same time, you have a problem," Laliberte said. That is where SRM tools can help. What you want to do, he noted, is see beyond the raw capacity to what has been allocated, mounted and used. SRM tools can also show you what the application thinks you have and what you truly have, so that as the physical infrastructure fills up you can add more capacity before a shortage develops. It's true that thin provisioning tools can provide much of this capability, but SRM tools can provide a holistic view, especially in an environment with multiple vendors, according to Laliberte. What you should know before you invest in SRM tools So where should a data storage manager start when considering SRM tools? John Webster, senior partner at the Evaluator Group, issued this warning: Be sure to make a distinction between device management tools and higher-level SRM tools that serve as "managers of managers." Device managers for specific arrays -- often provided by vendors -- can be helpful for provisioning, but they don't offer the broader view that most heterogeneous environments demand. "A higher-level SRM tool can potentially look across those arrays and take measurements and provide statistics on how capacity is being used and by whom," Webster said, "along with clarifying bandwidth issues and impact on devices such as host bus adapters." Webster said data storage managers have to understand where they land on the capacity-usage curve, as well as what impact carving out a piece of physical storage and bringing it online will have on the entire network. Irwin Teodoro, director of systems integration at systems integration firm Laurus Technologies, spent several years at storage vendors such as EMC Corp. He's also developed a degree of caution toward SRM tools. For starters, they can be pricey. Moreover, "they are so complicated" they sometimes aren't worth the investment, Teodoro said. "I would compare SRM applications to the iPhone, which has so many apps available but most people really just need to make phone calls and get their messages," he said. "We encourage our customers to look at the tools they already have before deciding to invest in SRM; often, they find they already have the functionality they really need." In fact, he added, unless your environment is very heterogeneous you may not need SRM at all. If you do decide to invest in SRM, Teodoro said, it's important to look for tools that handle historical data easily from multiple sources and "interfaces that are intuitive." Looking ahead, Webster sees higher levels of automation as inevitable, what one of his clients terms a "storage ATM" approach. "I think there's movement toward that in the industry because it would relieve storage administrators from this repetitive provisioning process," he added. Administrators would like to automate that process even to the point where a database administrator can have a portal and, based on answers to a set of questions, be able to activate additional storage on their own. "There are vendors now talking about delivering that kind of capability," he said. SRM tools landscape Here's a sampling of the wide range of SRM tools currently on the market: CA Storage Resource Manager and CA Vantage Storage Resource Manager (mainframe) BIO: Alan Earls is a Boston-area freelance writer focused on business and technology, particularly data storage. SRM tools offer capacity reporting, including change management and chargeback, but can be pricey and complicated. Here's what you should know before you invest in SRM tools. What you'll learn: As the size of data stores continue to grow rapidly, the challenge of managing data storage capacity on RAID is increasing. While this raises some questions about the integrity of RAID, new technologies are emerging with higher capacities and more capable drives.
Managing increasing amounts of data storage capacity on RAID has become a difficult task for many storage administrators. The massive amounts of capacity on a single drive also raise questions about the efficacy of RAID 5 as a data protection technique. For industry analyst Mike Karp, the growing capacity of multiple terabyte disk drives inevitably increases disk errors. "For every unit of capacity, we know there will be a certain number of errors," he said. Organizations traditionally relied on RAID 5 to overcome those errors through the use of parity data. However, because of the time it takes a system to rebuild a 1 TB drive using RAID 5, the likelihood that another drive may fail increases, compounding the problems. Karp's solution is object-based RAID, which relies on small units of storage and allows the system to rebuild the drive in smaller increments. Others suggest RAID variations that use multiple parity disks, such as RAID 6, can do the job. There are also a handful of new technologies emerging to address the capacity issues of data storage media. New technologies increase disk's data storage capacity Several technologies are vying to produce the next breakthrough in packing more bits on a piece of storage media. Currently, perpendicular magnetic recording (PMR), which stacks bits vertically on the surface rather than laid out horizontally, is the technology of choice, but it will hit the superparamagnetic limit within a few years. But alternative technologies are emerging, including: Bit-patterned media (BPM) stores each bit as a nanometer-scale pattern of grains on the media. As described by Hitachi, it creates an ordered array of highly uniform islands, each island capable of storing an individual bit. HAMR will likely be ready for production by 2013, with MAMR probably ready a year later, according to Mark Nossokoff, a senior member of the strategic planning team at LSI Corp.'s Engenio Storage Group. BIO Alan Radding is a freelance writer and frequent contributor to Storage magazine and SearchStorage.com. With data growth increasing every day, the challenge of managing data storage capacity on RAID is also increasing. This is raising some questions about the integrity of RAID, so new technologies are emerging with higher capacities and more capable drives. What you'll learn: I/O virtualization can help storage managers optimize their data center operations. This tip examines the options and best practices for initiating I/O virtualization to reduce bottlenecks and the physical infrastructure of your data center.
I/O virtualization, like server virtualization, adds an abstraction layer to simplify and optimize data center operations. In the case of I/O virtualization, there's an abstraction between the servers accessing interface cards and the actual cards themselves. The goal is to be able to share those cards across multiple servers. The technology counts on the assumption that most data center servers can't utilize network interface cards (NICs) to their maximum capabilities at all times. I/O virtualization attempts to better utilize the available bandwidth by allowing more servers access to each individual card. It's important to note that I/O virtualization doesn't provide more bandwidth to the servers, it just ensures that more of the available bandwidth is used. Features like quality of service (QoS) and N_Port ID Virtualization (NPIV) can help ensure that critical applications are getting a guaranteed level of performance. I/O virtualization products Let's take a look at some of the I/O virtualization products on the market today: Aprius Inc. Aprius G80 I/O Gateway Three ways to share cards in an I/O virtualization system The first step in initiating I/O virtualization is to understand what methods these systems use to virtualize NICs and host bus adapters (HBAs). There are currently three ways to share each card that's placed inside an I/O virtualization system. The first approach is one where individual servers take turns using an interface card. This approach provides value when it comes to the use of expensive application-specific cards that are only used at certain points of the day. A second approach is to use multiple-port cards that allow each port to be individually addressed. While this doesn't increase bandwidth utilization, because each server has its own port and bandwidth, it does reduce costs. For example, a quad-port card is typically less expensive per port than purchasing four individual cards. The challenge, however, is that most servers can't take advantage of that many I/O ports. I/O virtualization solves that problem by sharing the card on a per-port basis, leveraging the cost savings of the multi-port card while allowing each port to improve, but not be fully utilized. The final and most ideal option is to select NIC cards that support Single Root I/O Virtualization (SR-IOV). These cards have the intelligence to enable multiple hosts to share a single card and are ideal for I/O virtualization systems. There are a few 10 Gigabit Ethernet (10 GbE) cards available and most of the next generation of Fibre Channel over Ethernet (FCoE) cards will support this standard. Depending on the system, a card with SR-IOV should be able to share the bandwidth of that card selectively across multiple servers. SR-IOV cards in an I/O virtualization system work under the assumption that different servers will have peak needs at different times. If two servers need significant I/O resources and are using 6 Gb of a 10 Gb segment, the remaining servers are only using a fraction of that and can comfortably service I/O needs. Some I/O virtualization systems have the ability to spill over I/O to a spare card in the event that bandwidth utilization needs saturate the throughput of one card. Three ways to connect NICs to servers: Ethernet, InfiniBand and PCIe After selecting the cards that will be shared, the next step is to select the connection method to the servers that will have access to the I/O virtualization system. Presently, there are three competing methods to accomplish this; Ethernet, InfiniBand and PCIe. At first glance, PCIe seems to be the most natural fit given that PCIe cards are being shared. However, PCIe wasn't truly designed as a networking standard outside of the confines of a physical server. InfiniBand and Ethernet were both designed to be networked, but weren't designed to transport PCIe traffic. InfiniBand has the performance capabilities to support PCIe bandwidth requirements but its adoption rate, other than in back-end interconnects, has been relatively low. Ethernet, on the other hand, is ubiquitous and very networkable. For it to carry PCIe traffic today requires special logic be added to an Ethernet card. In the future, this capability could be built into standard Ethernet cards. Selection of the connection criteria needs to be done with a careful examination of each method. The organization needs to decide which method will provide the performance needed, the networking scalability needed and the familiarity of what is already in use in the organization. While cost varies greatly, the PCIe implementation should be the least-expensive option with Ethernet as a close second. How to determine which servers to include in your I/O virtualization system The final step is a determination of which servers would be the best possible candidates for inclusion into the I/O virtualization system. With the first two sharing methods-- one card one server or one port one server -- the bandwidth to the server is locked in. In the shared card method with SR-IOV, the biggest concern is that a server could starve out the other servers based on its I/O resource consumption. While some of the systems can prevent that from happening, it could mean that an individual server wouldn't be given access to the bandwidth it needs to keep the other servers sustained. The best practice is to identify the few servers in the environment that may have these high demands and maintain a separate connection to them. You could also leverage a standby card and let I/O needs spill over as mentioned above. Most I/O virtualization systems will let you dynamically assign a specific card to a specific server in the event you know there's a performance spike on the horizon. One of the safest ways to start with I/O virtualization is to use it as a repository for redundant cards, since most servers have redundant network and storage connections. In a 10-server rack that can mean up to 20 extra cards, which could add up to $40,000 to provide I/O redundancy in some cases. A starting point for I/O virtualization is to then move one or two of these redundant cards to the system and then not purchase secondary cards for new servers. This can be done by simply mapping the secondary connection to the cards in the I/O virtualization system. Typically, when a network or storage connection fails, all of the cards don't fail in all the servers. Instead, one card, a cable connection or an SFP connector on the switch fails. That one server could then map directly to the card in the system and continue operating until the primary card is replaced. BIO: George Crump is the lead analyst at Storage Switzerland, an IT analyst firm focused on the storage and virtualization segments. We examine I/O virtualization products; how I/O virtualization systems virtualize NICs and HBAs; and using Ethernet, InfiniBand or PCIe to connect NICs to servers. |