|Home > Data Centres > News > Latest developments for Fibre Channel standards + NAS Q and A||
SNS Europe Data storage and IT management: Data centre solutions, data centre solution products, data center solution
Separating Ethernet networks and FC SAN is a convenient solution for guaranteeing optimal provisioning of Fibre Channel storage-based services and Ethernet-based services. For many data center executives, maintaining separate physical networks for SANs and LANs continue to be the preferred strategy for some or all of their data center network applications now and in the foreseeable future.
But with the recent advent of FCoE (Fibre Channel over Ethernet), IT executives now have the option of consolidating their tried and true Fibre Channel storage-based services AND their tried and true Ethernet-based services onto a converged DCB (Data Center Bridging) 10FCoE Ethernet network.
To address increasing user demand for higher bandwidth there have been many exciting Fibre Channel developments in the Technical Committee T11 over the past several months. These include enhancements to the FCoE protocol, and the development of the 16GFC standard.
Additionally, there have been new standards developed in areas such as Simplified Configuration Management (FC-SCM), Inter-Fabric Routing, Energy Efficiency, and Switch Architecture. This article will discuss the current status of today’s published technical standards and the future developments of Fibre Channel.
Recently Published T11 Standards
Three important Fibre Channel standards were recently published at INCITS (http://t11.incits.org/)
Enhancements to FCoE
FC-BB-6 is currently under development in T11. FC-BB-6 describes enhancements and new functionality for the FCoE protocol. This includes support for simple point-to-point configurations and end-to-end configurations. One of the outcomes of the BB-6 development is a renewed interest in alternative topologies for standard Fibre Channel Fabrics. This will result in a new FC-SW-6 project in T11.
16 Gigabit per second Fibre Channel (16GFC)
The FC-PI-5 and FC-FS-3 standards define 16GFC which is the fastest edge connection and single lane interface defined for the data center. T11 is currently putting the final touches on FC-PI-5 and FC-FS-3 and it will be forwarded to INCITS for further processing and publication in August. These standards define the 16GFC architecture and describe how speed negotiation takes place between the 8B/10B and 64B/66B encoding schemes. Automatic Speed negotiation, requiring no user intervention, is the key to providing backward compatibility with the prior Base2 speeds (4GFC and 8GFC) specified by the FCIA Fibre Channel Roadmap.
Simplified Configuration Management (SCM)
A new development in Fibre Channel is the FC-SCM effort. This project specifies the requirements for Fiber Channel support in simplified environments that are often associated with small to medium business (SMB) markets. The document focuses on behaviors associated with management tools, hosts, Fabrics, and Fibre Channel switches. The FC-SCM standard is resolving comments and will be forwarded to INCITS for further processing and publication by the end of this year.
The Fibre Channel Inter-Fabric Routing (FC-IFR) standard defines how devices on different Fabrics may communicate without merging the Fabrics together. This simplifies the management and configuration of large configurations that may result due to the consolidation of data centers. Inter-Fabric Routing allows this communication through the use of a new Fibre Channel entity known as a Fibre Channel router.
The Fibre Channel router, along with zoning enhancements, controls the manner in which devices discover one another and ultimately communicate. The FC-IFR standard is complete and has been forwarded to INCITS for further processing and publication by the end of this year.
Fibre Channel Port Model
Due to the prevalence of virtualized environments in today’s data centers, Fibre Channel is being enhanced to enable these environments. Historically Fibre Channel has provided virtualization mechanisms such as N_Port ID Virtualization, frame tagging, and Virtual Fabrics to support SAN environments. In order to unify the Fibre Channel virtualization methods, an updated Fibre Channel port model was introduced in FC-FS-3 and FC-SW-5.
Prior to the port model update, the basic link level functionality was defined as the FC-2 level. To accommodate different virtualized environments, the FC-2 level was divided into three new sublevels. The Level 2 sublevels are the Physical (FC-2P), Multiplexer (FC-2M) and Virtual (FC-2V) sublevels . The FC-2P sublevel defines low level functions such as frame transmission and reception, and buffer-to-buffer flow control. The FC-2M sublevel includes the specification of addressing and routing functions. Finally the FC-2V sublevel presents the environment necessary to support multiple high level FC-4 mappings such as SCSI and FICON. By dividing the FC-2 level into three unique levels, the FCoE environment defined in FC-BB-5 is accommodated as well. The FCoE Entity simply replaces the FC-2P and FC-2M levels while providing the proper functionality to the FC-2V level.
Future Fibre Channel Developments
T11 is continuing to develop aspects of Fibre Channel to protect customer investment and bring Fibre Channel based solutions into new markets. These advances include higher speeds, advanced protocols, energy efficiency, and FCoE refinements. Here are some specific examples:
The FC-PI-6 project was accepted in June 2010 to define 32 Gigabit per second Fibre Channel (32GFC). The FC-SW-6 project was accepted in June to define additional Fibre Channel Switch functions to support advanced Fabric topologies. These advances include support for Fibre Channel Fabric and FCoE integration, and the possible extension of FSPF for distributed FC Forwarding topologies. On the management side, FC-GS-7 project has been created to provide Fiber Channel service enhancements for FCoE and Fibre Channel Fabric environments.
T11 is also discussing the task of defining requirements for energy efficient Fiber Channel to address the green aspects of data center design and implementation. This effort will likely result in one or more additional Fibre Channel projects being created in the next few months.
Fibre Channel standards continue to evolve to meet the needs of ever-expanding SANs. Several projects are just completing including the standards for 16GFC, Inter-Fabric Routing and Simplified Configuration Management. The Fibre Channel standards have been enhanced to improve the adaptability of Fibre Channel to virtual environments and FCoE. Work has begun to enhance the basic Fabric models to accommodate new topologies in FC-SW-6 and management schemes in FC-GS-7. Work has also begun on 32GFC in FC-PI-6 to ensure that Fibre Channel remains as the fastest single-lane interface in the data center. Additionally, IEEE standards are emerging in the area of Data Center Bridging (a must for FCoE transport), and increased physical layer Ethernet switch-to-switch bandwidth for core inter-switch links (ISL) in the form of 40GFCoE and 100GFCoE ISL’s.
(This article acknowledges the following contributors: Steve Wilson, INCITS TC T11 Chairman, Brocade Herbert Bild, SNIA Europe German Country Committee member, NetApp Gilles Chekroun, SNIA Europe Board of Directors, Cisco Systems).
NAS Q&A session
What are the criteria for choosing a NAS system today?
NAS systems are appropriate for a variety of workloads and applications; from home directories, to virtual desktop infrastructures, to database and business applications; to the storage and analysis of large quantities of seismic data. They can provide Windows CIFS (SMB) protocol, or NFS, or both, and from handfuls of disks to multiple petabytes in scalable systems. Criteria common to all of these is a good match of business need to the proposed system; capacity, performance, price point, required reliability and availability. As a result of the recent global macro economics, more purchasing decisions are requiring substantiated return on investment metrics, including cost per TB, price per performance metric, future expansion capability and IT management cost. NAS compares very well on these metrics and consequently has gained market share through the recession. Important too is the implementation and adherence to industry standards, from NAS protocols to rack sizes.
Do entry-level NAS systems (models with two to eight hard disks) already lower operational expenditure? Or are they all about taking the storage-burden off a PC-based server?
Both. An appropriate entry level NAS systems can reduce data storage costs, simplify management and provide enhanced levels of performance and protection over a PC based server based solution. Facilities available from NAS storage systems, such as snapshot based backups, compression of data and so on are additional advantages that are often not available or easily implemented on PC based servers.
In many enterprises, NAS systems are used for secondary storage only. Under which use scenarios can they function as primary storage? What need IT-administrators be aware of when they are looking for a NAS system that is to function as primary storage?
In many cases, NAS systems, particularly those offering NFS protocols, are more effective as primary storage for many key business applications, such as databases, virtual desktop infrastructures, and Web2.0 applications. Very often, these applications are cheaper and easier to implement and manage using NAS than their block-based SAN equivalents.
Buyers should use the same criteria as they would for block based SAN: will it perform? does it adequately protect my data? is it expandable should there be growth? and does the application vendor support the NAS implementation of the product satisfactorily?
Are combined NAS/SAN environments a trend for the enterprise segment, are they here to stay? Or is it just a stopgap measure to combat the current data explosion?
In our opinion, they are here to stay. The flexibility they offer removes several issues associated with separate NAS and SAN systems. Capacity can be better utilized by sharing the same pool of disks. With modern unified systems, performance is not compromised; in fact, it may be improved due to combining separate SAN and NAS disk counts into a larger pool of disks for both. With a move to virtualization and cloud storage, which tend to benefit more from NAS than from SAN, enterprises can fix today’s problems and have the flexibility to cope with tomorrow requirements. Unified systems avoid the need and eliminate the overheads of silos of storage.
What technology developments can CIOs and IT planners expect in the next two to three years regarding NAS?
Virtualization and scalability are the two big upcoming trends in which NAS will make the biggest impact. From a connectivity perspective, the cost reduction of 10Gb Ethernet, 10Gbe connectivity down on the client motherboard, and link aggregation of four 10Gbe ports deliver the potential for up to 5GB/sec bandwidth. The advent of 40Gb Ethernet switches and storage systems supporting NFSv4.1 (which includes parallel NFS) will deliver scale-out capabilities not available with any storage solutions today, as well as a native feature set including the integration of the management communication infrastructure into the file system with file state change awareness and meta data attribute support. Add to that Hypervisor awareness of snapshots, replication and data deduplication and you get an enormously strong storage value proposition for cloud and virtual server environments.
The answers were provided by the co-chairs of the SNIA’s Ethernet Storage Forum NFS Special Interest Group (SIG), Alex McDonald, NetApp and Don Grabski, Panasas.
Tags: Data Centres
|Related White Papers|
|Read more News »|
|Related Web Exclusives|
|Related Magazine Articles|
|White Paper Downloads|
Keep up to date with the latest industry products, services and technologies from the world's leading IT companies.