8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? Version 1.0 As you are well aware, Fibre Channel technology is changing rapidly, especially in the FICON and SAN switching arena. This is really great news as it provides customers with alternative choices for their mainframe I/O infrastructures rather than having to buy a one size fits all solution. And Brocade is once again the leading innovator in I/O infrastructure connectivity.
INTRODUCTION There is much discussion today, as there always has been in the past, about the future of Fibre Channel in the enterprise data center compared to other emerging technologies. That discussion, very often, seems to be pushed along more from vendor hype and claims than from actual customer needs and requirements. In fact, in September 2011 Gartner, Inc. analysts made an interesting update to their IT Market Clock series specifically for the Storage Technology market. What they show is that Fibre Channel Networking has just reached the Zenith of Industrialization of the technology lifecycle. Gartner s IT Market Clock for storage technologies http://www.gartner.com/id=1784215 This series of reports highlights the progress of IT products and services from the time they first become viable to deploy and use, to the time when they must be retired. In this report, Gartner says Fibre Channel has just reached the Zenith of Industrialization with a time horizon of five to 10 years before it reaches the next phase of the lifecycle. [September 2011] What Gartner means by Zenith of Industrialization is that Fibre Channel has become a mature, proven technology that has demonstrated its utility and cost benefits. Below is a Gartner statement to end-users about how they should view this technology: Users should continue to view FC-based SANs as the mature, high-availability, large-scale option for SAN deployments for the next several years. For high-performance applications that require scalability and high levels of reliability, deterministic performance and several highly available simultaneously active storage paths, FC is a cost-effective and proven technology. 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 2 of 8
End-users get this, of course, which is why the vast majority of our end-user base have told us that they will continue to invest, if not increase their investment, in Fibre Channel switching technology as per this data point from our annual customer survey in 2011: Fibre Channel Purchase Plans Strong and Growing Results of the Brocade Annual Customer Survey Survey question: How do you expect your company s Fibre Channel purchases to change in the next 3 years? Don't Know 9% Will Decrease 12% Will Increase 43% No Change 36% n=503 ~80% of Brocade customers expect to maintain or increase Fibre Channel spending over the next three years Source: Brocade Annual Customer Survey, July 2011 Mainframe customers positioning themselves for a pending I/O infrastructure technical refresh now have a real selection of FC products from which to choose in order to meet the ongoing strategic and tactical requirements of their business. Both 8Gbps and 16Gbps Directors and modular switches are available from Brocade to make it easier for a customer or prospect to match their requirements against an appropriate product set from Brocade. Of course Brocade s most current FICON technology is our flagship DCX 8510-8 and DCX 8510-4 Directors (aka Backbones when reading Brocade literature) as well as our Brocade 6510 modular switch. These products provide up to 16Gbps of non-blocking FC port link speed along with high availability and great scalability attributes. Brocade s DCX and DCX-4S Directors and the B5100 and B5300 modular switches provide customers with our very mature 8Gbps non-blocking I/O architecture which have been the most popular switching products deployed in mainframe data centers since 2008. HAVING A CHOICE Of course, the problem with having a choice is in making a decision. One must have some criteria in mind in order to select an appropriate platform upon which you will spend budget and then additional resources like manpower and time on design and deployment. In an effort to be helpful to you, provided below are some of the benefits and considerations that you might be most interested in when making the choice between 8Gbps or 16Gbps I/O infrastructures for your data center. LET S START WITH 8GBPS Investment protection is one of the great capabilities that our 8Gbps DCX family of products provides to you. Brocade will provide you with an upgrade path from 8Gbps DCX Directors to fully functional 16Gbps platforms. If you need to deploy the more economical 8Gbps technology today then you will not be shut out from upgrading those acquired assets to 16Gbps as your needs increase. If you have Brocade 4Gbps FICON modular switches, for example to front-end your physical IBM and/or Oracle tape drives, those FICON switches can attach to 8Gbps FICON Directors through ISLs which will automatically autonegotiate to be 4Gbps switch-to-switch links. This is fully qualified and supported by IBM for FICON deployments. 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 3 of 8
You might want to use our 8Gbps DCX family if you have a need to interoperate, in the same fabric, with older McDATA FICON Directors. If you have old McDATA 6140 or I10K FICON Directors then they have been qualified by IBM to interoperate with our 8Gbps DCX FICON Directors when running our Fabric OS (FOS) firmware at level 6.4.2a. There is not any IBM qualification nor support for interoperability between old McDATA Directors and the 8Gbps DCX- 4S. If you are considering deploying an I/O infrastructure where you will have multiple operating systems being hosted on switching devices (SAN Only like Windows, AIX, Linux, Unix, etc. or even FICON along with one or more other operating environments) then that is known as a Protocol Intemix Mode (PIM) environment. To provide the highest availability and data security you should always utilize zoning, virtual fatrics and port prohibits or port blocking of some sort when deploying a PIM infrastructure. There are a number of ways to turn ports off so that accidental miscabling does not cause any data center problems. And there is one good way to keep ports on but not allow them to connect with any other ports on that device -- that is known as port prohibiting. Port prohibits can be provided by a facility known as Prohibit Dynamic Connectivity Mask (PDCM) through the Allow/Prohibit Matrix. The Allow/Prohibit Matrix becomes available to users when the control function of FMS is enabled through CLI or with DCFM/BNA. At 8Gbps, enabling FMS and using PDCM prohibits to prevent accidental mis-cabling between operating environments by inhibiting port communications (port prohibits) for any port on the chassis that does not have a cable plugged into it, is a no cost feature. Be aware that changes were made to FOS 7.X that allowed FMS to be enabled only if a Control Unit Port (CUP) license is also on the chassis. The CUP license is a chargeable feature. So with 16Gbps platforms (or upgrading to FOS 7 to manage 8Gbps platforms) utilizing PDCMs to prohibit ports against accidental mis-cabling will require the purchase of a FICON Management Server (CUP) license. If you have any 1Gbps storage to which your I/O infrastructure must attach then the 8Gbps DCX family are the correct platforms for you. Deploy 4Gbps optics on the 8Gbps physical ports that will attach to the 1Gbps storage. Optical transceivers (SFPs) can only negotiate three speeds. A 4Gbps optic can attach to 1Gbps, 2Gbps and 4Gbps storage and host connections. If you currently use a mainframe with only FICON Express2 channel cards and/or FICON Express4 channel cards then you might find our 8Gbps FICON DCX Directors will provide you with the best performance and scalability at an economical price in that low speed environment. If you have recently obtained a mainframe that uses FICON Express8 channel cards and CHPIDs, and those channel paths are only moderately busy, then you might find that our 8Gbps FICON DCX Directors will provide you with nonblocking performance and superb scalability at an economical price in that low-to-moderate I/O workload environment. When switched-ficon is deployed it automatically makes use of Fan In Fan Out. By optimizing Fan In Fan Out the customer can ensure that all of their CHPIDs and storage ports are working at optimal levels and providing the highest possible return in value for the cost of those assets. If cascaded FICON is not being deployed then our DCX 8Gbps Director products will allow you to provision the best possible high speed, non-performance blocking FICON infrastructure to connect your 8Gbps mainframe CHPIDs with your 8G, 4G and 2G storage devices. If you are using FICON Express4 on the mainframe and 4Gbps storage, DCX can provide you with 384 ports (and the DCX-4S with 192 ports) of 4Gbps throughput without any performance blocking and with plug-and-play ease of use. Either the 8Gbps ports will auto-negotiate to 4Gbps or the customer can deploy 4Gbps optics in the 8Gbps physical ports. If you are using FICON Express8/8S and 8Gbps storage, DCX can provide you with 256 ports (and the DCX-4S with 128 ports) of 8Gbps throughput without any performance blocking and with plug-and-play ease of use. If you are using FICON Express8/8S and 8Gbps storage, and you require more than 256 ports of local connectivity within a FICON Director chassis, then we can provide you with two options: Option 1: You can deploy Inter-Chassis Links between two DCX chassis s, each with up to 384 ports of connectivity. ICLs interconnect multiple chassis together without having to use up valuable channel and control unit ports. With ICL s, every FC port is available for device connectivity. And with up to 512Gbps of ICL bandwidth available between a pair of Brocade DCX chassis s you can ensure a cost-efficient, 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 4 of 8
Option2: highly scalable design. Competitive designs must use front-end director ports for interconnectivity, wasting 25-35 percent of the FC ports and using 8x more cable for ISL connectivity. A DCX can provide you with 384 ports of 8Gbps throughput (using 48-port blades) without any performance blocking although it will take a little more management for you to deploy our innovative Local Switching along with central switching for complete chassis non-blocking performance. Combine any of these options with a very low per frame latency of 2.1 microseconds and ultra fast Cut-Through frame routing and you have the best 8Gbps I/O infrastructure that it is possible to deploy. It is difficult for a layman to get a handle on how much Linux is running on System z. Research suggests that by the end of 2011 about 64 out of the top 100 mainframe clients in the world run Linux and that Linux on System z accounts for approximately 19 percent of the total mainframe MIPS installed worldwide. Although unverified, it appears that 30 to 35 percent of IBM System z customers are actively using Integrated Facilities for Linux (IFLs) specialty engines. So it seems accurate to state that Linux is big business on System z. If you are one of the many customers running Linux on System z, and you continue to use SCSI data files, then Node_Port ID Virtualization (NPIV) is what allows you to consolidate down to the fewest number of FCP CHPIDs possible while at the same time providing each Linux guest with abundant performance. CHPID consolidation and exceptional performance are both possible while also achieving high utilization levels on each of the mainframe FCP paths. NPIV can only be utilized for Linux on System z when switching devices are used to connect together the customer s I/O infrastructure. 8Gbps FCP links allow more guests to utilize a virtualized (NPIV) single physical path. To take full advantage of Linux on System z customers should be deploying FICON Express8 or FICON Express8S CHPIDs in conjunction with DCX or DCX-4S 8Gbps connectivity ports. NOW LET S CONSIDER A 16GBPS I/O INFRASTRUCTURE Just as it was true on our 8Gbps Directors, investment protection is one of the great capabilities that our 16Gbps DCX family of products provides to you. If you already own an 8Gbps DCX and/or DCX-4S then Brocade will provide you with an upgrade path from those 8Gbps DCX Directors to fully functional 16Gbps platforms. The asset that you have already purchased will now have an extended lifetime of usage. The 16Gbps DCX 8510-8 can provide you with 384 ports (and the DCX 8510-4 with 192 ports) of 16Gbps throughput without any performance blocking and with plug-and-play ease of use. Combine that with a very low per frame latency of 2.1 microseconds and cut-through frame routing and you have the best I/O infrastructure that it is possible to deploy. When IBM provides 16Gbps CHPIDs then that same 16Gbps DCX 8510-8 will be able to provide you with 256 ports (and the DCX 8510-4 with 128 ports) of 16Gbps throughput without any performance blocking and with plug-andplay ease of use. Combine that with a very low per frame latency of 2.1 microseconds and cut-through frame routing and you have the best 16Gbps I/O infrastructure that it is possible to deploy. Inter-Chassis Link (ICL) connectivity is also available on the DCX 8510 Director family. On the DCX 8510, there are up to 2 Tbps of ICL bandwidth available and a new 50 meter universal multi-mode optical cable, so that you can ensure that you have a cost-efficient, very flexible and highly scalable design. Gartner, Inc. recently published a paper on planning for the next generation of IBM Mainframes. http://www.gartner.com/id=1891914 In this paper they describe the history of IBM System z mainframe announcements and they predict that it probably will not be too long before IBM announces a new offering in the mainframe arena. Although only IBM knows for sure, it is possible that any new mainframe offering will, at some point, be capable of provisioning channel CHPIDs at 16Gbps. After all, IBM has already qualified the Brocade 16Gbps DCX family for FICON I/O connectivity so they are certainly aware that 16Gbps FICON connectivity is available and that storage vendors, including themselves, will announce 16Gbps attachment to storage as soon as that becomes strategic for them to provide. At the point in time that IBM provides 16Gbps CHPID capability, customers will need to match that with Brocade s DCX 8510 16Gbps connectivity ports. Brocade has dramatically enhanced cascaded links (ISLs) when utilizing Brocade s 16Gbps product set. Featured below are enhanced or new capabilities offered through our 16Gbps technology. 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 5 of 8
A significant new capability for 16Gbps FICON F_Port and ISL links is that the data encoding on the frame is now 64b/66b with only a 2% overhead compared to the 8b/10b data encoding that is used for 1-8Gbps at 20% overhead. Frames sent across the F_Ports and ISL links will be packed with more data providing additional efficiency and effectiveness for the entire I/O infrastructure. Often customers find themselves deploying cascaded FICON. Cascaded FICON can be implemented through the use of dark fibre runs, xwdm and FCIP. When using dark fibre cascaded FICON (single mode FC cables running between two switching devices usually in two different data centers) the faster the link speed on those long distance links the better the overall I/O performance will be. DCX 8510 can easily host many 16Gbps Inter-switch Links (ISLs) to provision the best possible high speed, non-performance blocking FICON storage network. When customers are using DWDM for deploying cascaded FICON there are significant charges for the bandwidth that is deployed between the two sites. It is more cost effective to utilize one 10Gbps DWDM link than to use multiple but slower 1 or 2Gbps DWDM links. On our 16Gbps switching devices the customer can configure a 16Gbps port to run at 10Gbps and then buy a 10Gbps SFP+ to deploy into that port. That 10Gbps port can then attach to the 10Gbps DWDM port providing you with significant Telco bandwidth savings. And for DCX 8510 connectivity blades, Brocade has increased buffer credits per Condor3 ASIC to 8,192 per 16-port group on 32-port blades which is 4X more than was available at 8Gbps. Besides having more buffer credits for long distance FC connectivity we have new capabilities to recover lost buffer credits. FOS v7.0 adds support for the detection and automatic recovery of buffer credits on 16G capable ISLs (both ends of the link must be terminating on 16G capable port). Loss of a single buffer credit is detected at individual virtual channel (VC) level and is automatically recovered without performing link reset. Loss of multiple buffer credits and stuck VC (VC in permanent zero buffer credit condition) conditions are recovered by performing link reset. Additionally, there is now a new Inter-switch Link (ISL) testing capability that customers have been needing for years and that should soon be qualified for FICON. Since the advent of switched fibre channel fabrics, when a fiber cable is connected between two switching devices and activated it immediately became an E_Port (ISL) without any real capability to thoroughly test it before using it. Thankfully, now that has changed for SAN and will soon change for FICON. On Brocade 16Gbps platforms a D-Port (Diagnostics Port) is a special port mode that an eventual ISL port on Brocade's 16G platforms can be configured into first. When deploying cascaded FICON, what we now recommend is that a customer should configure the ports on each switching device that will be attached together to form ISL links as D_Ports, and get those links thoroughly tested, before allowing them to be deployed as E_Ports. Having Diagnostics ports on both sides of the link will allow a customer to test their connections between two switches without actually deploying working ISLs. This means there will be no user traffic and also no fabric management over this link and so there will be no impact to the fabric at all. From a fabric perspective, the ISL has not been deployed and is not active. A connected pair of D_Ports will be tested using several automatic tests: Either 16Gbps ports or 10Gbps ports can become D_Ports. However, 10Gbps ports cannot participate in the Electrical Loopback tests although it does participate in the others: Electrical loopback - (only with 16G SFP+) tests the ASIC to SFP connection locally Optical loopback - (with 16G SFP+ and 10G SFP+) tests the whole connection physically. Link traffic test - (with 16G SFP+ and 10G SFP+) does latency and cable length calculation and stress tests Forward Error Correction (FEC), available on 10Gbps and 16Gbps ISL links, is another new capability of the Brocade 16Gbps switching devices. FEC is a system of error control for data transmissions, whereby the sender adds systematically generated error-correcting code (ECC) to its transmission. This allows the receiver to detect and correct errors without the need to ask the sender for additional data. Our 16Gbps switching devices can fix up to 11 bit errors per 2112 bits of payload transmission (11 errors corrected per every 264 bytes of data in a frame) on a 10Gbps or 16Gbps ISL port. If you have bit error problems on your current ISL links (FC, DWDM or FCIP) then FEC can make a significant difference by improving the performance across those error prone links. In some Proof of Concept (POC) tests, on a very dirty cable link, we were able to use FEC to completely clean up all of the bit errors that had been previously generated. FEC can significantly increase performance on long distance links. 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 6 of 8
Historically, once an ISL connection is established and working, it has been very difficult to remove a misbehaving E_Port connection. When using FOS 7.0 and higher, the new Port Decommission feature provides customers with the ability to non-disruptively remove any ISL from service. When an ISL is selected for decommissioning, the switches communicate with each other to coordinate the movement of frame flows off of the targeted ISL to alternative ISL paths. Once the frame flows are moved to those alternative paths, the switches block the E_Ports associated with the ISL being decommissioned to complete the decommission process. YOU WILL BENEFIT FROM SOME BROCADE FEATURES WHETHER YOU CHOSE AN 8GBPS OR A 16GBPS PLATFORM EXPERIENCE AND DEEP EXPERTISE Brocade s FICON experience and deep networking expertise is unmatched. Brocade has been in the mainframe storage networking business for more than two decades, going back to parallel channel extension technology in the late 1980s. In addition to pioneering ESCON channel extension emulation technology to enable superior performance over long distances for business continuity/disaster recovery (BC/DR), Brocade manufactured the 9032-3 and 9032-5 ESCON directors for IBM. Brocade worked closely with IBM to develop FICON technology, and delivered a complete portfolio of industry leading FICON directors and switches, increasing the performance, reliability and availability features at every step. As a result, Brocade has several patents and shares patents with IBM on technologies such as the FICON bridge card and Control Unit Port (CUP). Brocade also launched the industry s first FICON certification and accreditation program, a top rated program that even certifies IBM personnel. Today, Brocade continues our heritage of innovation and mainframe storage networking leadership with both the 8Gbps Brocade DCX Director family and the industry first 16 Gbps DCX 8510 Director family, a full portfolio of 8 and 16 Gbps FICON switches, and FICON channel extension with emulation technology, enabling new FICON applications over any distance. BEST ENERGY EFFICIENCY Brocade products use fewer active components in order to drive energy usage down while maximizing throughput speed. Because of our efficient product design, our products use as little as 1/10th the power of our competition. In fact, independent testing demonstrated that a fully populated 384 port 8Gbps DCX backbone consumes less power than the competitor s director when they have NO port modules installed at all. Brocade delivers the highest energy efficiency, and that translates to real operational savings. HIGHEST RELIABILITY AND AVAILABILITY The Brocade DCX again leads the way by a wide margin over the competition when considering Reliability, Availability and Serviceability (RAS). With millions of ports deployed in mission critical applications in all major verticals around the world, Brocade s storage OEM partners consistently rank Brocade products #1 in terms of reliability, having achieved well over five nines of availability in field proven applications. EASE OF DEPLOYMENT AND MANAGEMENT FICON can be easily deployed with Brocade DCX Backbone directors, using simple zoning and z/os HCD definitions. Brocade Network Advisor (formerly called Brocade Data Center Fabric Manager) offers centralized configuration management, performance monitoring and security tracking in a single pane of glass. Competitive offerings, in addition to HCD definitions, require ports be placed in a Virtual SAN (VSAN), a complex and time consuming exercise. Our competitor s VSANs, in a FICON environment, also restrict the number of ports available for host and storage connectivity. SUMMARY BROCADE IS YOUR BEST FICON PARTNER! With over 15 years of experience, Brocade is the only partner to work with IBM from the early Bus & Tag and ESCON days--to now delivering the industry s highest performing FICON directors. The Brocade DCX Director s superior performance, scalability, reliability, ease of use, and energy efficiency--as well as our extensive experience in mainframe storage networking-- make Brocade the best and safest choice for your mission critical mainframe applications. 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 7 of 8
For additional information please visit www.brocade.com 2012 Brocade Communications Systems, Inc. All Rights Reserved. 02/12 Brocade, the B-wing symbol, BigIron, DCX, Fabric OS, FastIron, IronPoint, IronShield, IronView, IronWare, JetCore, NetIron, SecureIron, ServerIron, StorageX, and TurboIron are registered trademarks, and DCFM, Extraordinary Networks, and SAN Health are trademarks of Brocade Communications Systems, Inc., in the United States and/or in other countries. All other brands, products, or service names are or may be trademarks or service marks of, and are used to identify, products or services of their respective owners.. Notice: This document is for informational purposes only and does not set forth any warranty, expressed or implied, concerning any equipment, equipment feature, or service offered or to be offered by Brocade. Brocade reserves the right to make changes to this document at any time, without notice, and assumes no responsibility for its use. This informational document describes features that may not be currently available. Contact a Brocade sales office for information on feature and product availability. Export of technical data contained in this document may require an export license from the United States government. 8Gbps FICON Directors or 16Gbps FICON Directors What s Right For You? 8 of 8