IBM System Storage ™ 存储概念: 入门I I would like to welcome you to one of two storage basics training. To ensure that you are providing accurate information and recommending the appropriate IBM systems to your customer, it is important that you have a basic understanding of storage. And that is what we hope to provide you in these two training sessions.

数据与磁盘 数据可以从磁盘写入或则读取 应用/ 文件系统采用数据块级别的 I/O 用户采用的是文件级别的I/O 典型地通过一个本地应用或是服务器应用 应用/ 文件系统采用数据块级别的 I/O 数据的收发是通过I/O 块 用户采用的是文件级别的I/O 收发的是数据文件 0 1 1 0 1 0 0 1 0 1 0 1 1 1 0 1 0 1 0 0 0 1 The underlying basis of storage and what information, or data, is written to and read from are the disk drives. As we go through these sessions, specifically in the second session, we will look at data from the application perspective and the different levels of data – and this can be either file-level or block-level data.

Part I结束时: 主机 光纤子系统 – DS3000 and DS4000 Fibre Channel Fibre Channel HBA 光纤通道 Controller Fibre Channel HBA 光纤通道 RAID Controller You are looking at the completion of part 1. By the end of part 1, you should understand the basics concepts such as HBA, what an HBA is, what FC is, and how FC is used to connect hosts directly to your disk subsystem. You will understand concepts such as DAS, RAID and controllers. In addition you should be able to list the components that build a complete storage subsystem. 两个独立控制器 带自动失效切换以获得持续的高可用性以防止一控制器或光纤通道失效 以太网到客户工作站

Part II 结束时: Servers A B Servers 存储区域网络 (SAN) 网络附加存储 (NAS) RAID Switch A B A B Switch 存储区域网络 (SAN) A B A B Switch Switch A B A B By the end of part 2 you will extend your knowledge beyond that of DAS configurations to be able to understand concepts such as a SAN, what makes up a SAN, and how your storage subsystems fit into a SAN. As well as concepts like Networked Attach Storage, also known as a NAS. We will finish up with a convergence of these two technologies and what does the convergence of a SAN and NAS really mean. Storage subsystem N A S N A S N A S A B Servers 网络附加存储 (NAS)

自测问题: 什么是存储? 什么是 DAS? SCSI 和SAS是什么和其用途? 什么是JBOD? 为什么我们需要使用RAID? 我是否理解 光纤通道Fibre Channel的优势? As you go through this first session, I would like you to ask yourself some specific questions. What is storage? What is DAS? What is SCSI and what is SAS? What is a JBOD? Ask yourself, why would I recommend RAID to a customer? And do I understand fiber channel and how it benefits my customer?

过去对于信息的观点 If you look at yesterday’s view of information, it was really a distributed view. You had islands of information separated typically by departments. For example, operations may have its own separate database, marketing might have its own and finance might have its own as well. This distributed view of information led to a concept known as DAS, or direct attach storage.

小测验 什么是DAS的定义 ? Dual Asynchronous Storage 双异步存储 Data Archival Software 数据存档软件 Direct Altered Storage 直接汇聚存储 Direct Attached Storage 直接连接存储 Now that DAS has been introduced to you, lets see if you have a true understanding of what the concept stands for. So I will give you a couple seconds to think about it, write it down, and decide what DAS stands for. Is it (a), Dual Asynchronous Storage, (b) Data Archival Software, (c) Direct Altered Storage, or (d) Direct Attach Storage.

测验答案 DAS的定义是 ? - Direct Attached Storage 直连存储 Well, I hope you came up with the correct answer, and that is (d) Direct Attached Storage.

直连存储 And lets look a little closer at DAS, or storage which is directly attached. And remember the earlier slide where we looked at islands of storage, or islands of information. Each one of these individual islands can have their own pool of information, or in effect, its own DAS device. That is the concept of Direct Attach Storage. DAS Device DAS Device

直连存储 (DAS) 存储直接连接到一台服务器上 内部磁盘 外部磁盘 SCSI, SAS, iSCSI, FC 块级别 I/O 具备/不具备RAID保护 外部磁盘 存储系统 基于控制器的RAID引擎 Direct attach storage has several different connection to host options as well as configurations. Storage can be attached via several different interfaces based on performance, reliability and cost. This includes SCSI, SAS, iSCSI and FC. For this training, we will focus on the two primary DAS interfaces – SCSI and SAS. All these host connectivity interfaces transmit and receive data with block level I/O. DAS can also be categorized as internal or external. Internal DAS, can be defined as drives housed within the server itself. These drives can be utilized with RAID or without. We will review RAID in greater detail later in this session. External DAS, on the other hand, is just that – external, in that the drives are housed in an enclosure separate from the server. Drives are housed similarly in an external JBOD, however, the RAID is managed within the storage system itself via controllers.

内部 DAS Application Server Application Server Application Server Motherboard Motherboard Motherboard Motherboard SW ROC HA 内部磁盘 无RAID 内部磁盘 软件 RAID 内部磁盘 主板集成RAID (ROC) 内部磁盘 RAID 主机卡 Lets take a closer look at internal DAS. The most basic variation of this is simply drives housed within the server. There is no RAID applied to these drives. For internal DAS with RAID, we can have several variations. One of these variations is RAID that has been applied with software that has been downloaded on the server. This is software based RAID. In the third example, we have ROMB, or RAID on the Mother board in which the RAID processing intelligence has been installed on the mother board. A final internal DAS based configuration is RAID applied via a PCI RAID card which is installed within the server via a PCI slot. And with an internal DAS, it is important to note that while typically the drives are housed within the server, in some cases they may external and reside on a JBOD. This would still be considered internal DAS because the RAID engine would still reside internal to the server in the various components I mentioned above.

外部DAS Application Server Motherboard HA Controllers 通过主机连接卡连接的外部存储系统 基于控制器的 RAID For External DAS, the RAID engine is on the controller which we discussed and is resident on the external array and connected via a host adapter. The drives are also housed with the same storage system as the controller. And as drives are added and the organization grows beyond this system, JBODs can be added to the system with the controllers on the first system managing the connected JBODs as well.

内部DAS 优势 外部DAS 优势 成本低 驱动器通常在服务器内部 具备一些数据保护 比将数据存储在客户端更好 增加的可靠性和性能 使用迅速 具备一些数据保护 外部DAS 优势 比将数据存储在客户端更好 增加的可靠性和性能 RAID保护 增强的特性和功能性 可扩展性 As you have familiarized yourselves with both of these configurations, the advantages are probably becoming quite clear to you. Internal DAS is first and foremost – low cost and likely an investment made when the server was bought and included these drives. And with drives and if RAID has been utilized, some data protection is included. External DAS offers the inherent benefit of being external – no reliance on the server and the possibility of going down. The external system will continue to be up, regardless of the server being online. With the various RAID algorithms which we will discuss later, such as mirroring, striping and parity, you have increased availability and performance. Many external RAID configurations provide additional features such as additional replication features for data mining and disk to disk backup. Scalability can be achieved by starting with several drives and a single controller and then adding additional drives and a second controller as capacity increases and availability becomes more important.

DAS缺点 每台服务器都有其自己的存储部件 ($$) 管理困难 备份很昂贵 浪费的存储空间 难于共享数据 降低了生产效率 It is also fair to say that DAS is not always the best implementation although it is a very commonly used configuration. Bottom line is, if each server has its own direct attach storage device, this can be an expensive. It is certainly expensive to manage all these individual DAS systems across the organization. It is also expensive to back up each of these systems to tape. It certainly wastes storage space with some systems being over-utilized while others are under-utilized. And it is difficult to share data – consider this, if I am in the marketing department and I need additional production space, it would be very difficult to know if there is additional capacity out there in other department storage subsystems that I could use. Even if I knew of this additional space available, the opportunity to utilize this available storage within a distributed DAS environment would be very difficult.

共享的DAS 存储直接连接到一台单独的服务器 存储系统包括RAID控制器和磁盘柜 SAS, iSCSI, FC RAID引擎是基于控制器的 A newer concept of DAS that reduces the disadvantages of distributed DAS is shared storage. As with the DAS concept we have been discussing, storage is directly attached to the server. However with shared DAS, more then one server can connect to a single storage system. Based on the number of host ports on the disk system, the number of controllers in the system, and whether availability is important to your customer, you can fit anywhere from 2 to 6 servers typically in a shared storage environment where these servers will house their data on a single storage system. This shared DAS concept can be achieved with SAS, iSCSI or FC host connectivity. This is also based on external disks systems which consist of RAID controllers in which the RAID engine is controller-based.

SCSI是什么? SAS 是什么? 微型计算机系统接口(SCSI) 串联 SCSI Pronounced “skuzzy” And with that, lets get into the two most commonly utilizes DAS interfaces. S – C – S – I stands for Small Computer System Interface. It is interesting acronym and is pronounced skuzzy for short. In addition to SCSI, we need to be familiar with a newer interface based off of SCSI. This is S – A – S, or SAS. SAS stands for Serial Attached SCSI. Outside of what these terms stand for and how they are pronounced, it is important to understand what SCSI and SAS actually are.

SCSI接口….started it all! 工业标准 I/O 总线 Standard connectors are the same on each device All devices share a common bus SCSI is in fact a high-performance standard, as well as an industry standard. And each device that connects via the SCSI interface all use standard connectors and share a common bus. You will also sometimes here SCSI termed as Ultra SCSI. This is more of a marketing term that defined the 320MB/s SCSI specification. Another important thing to note here, is that although there have been a lot of improvements in performance, as far as bandwidth is concerned with SCSI, your maximum length and number of devices is still relatively low. You are going to see with newer SAS technology that the cable lengths and device addressability is greatly improved and we will discuss this in greater detail later in this session. 8-bit data bus 16-bit data bus

哪些设备使用 SCSI? Disk Drives Tape Drives Removable Media Drives (Zip) CD-ROM Drives CD-R/CD-RW Drives Optical Memory Drives Media Changers And what type of devices use SCSI? Well a lot of devices do and not just disk drives. This includes tape drives, removable media such as Zip drives, and CD-ROMs .

小测验 SAS 是什么? Silicon Attached SCSI 硅连接SCSI Serial Attached SCSI 串联SCSI SCSI Attached Storage SCSI附加存储 Storage Attached Systems存储附加系统 And this leads us into our next interface – SAS . And with that, lets see if you know what SAS stands for. Is SAS an acronym for (a) Silicon Attached SCSI, (b) Serial Attached SCSI, (c) SCSI Attached Storage, or (d) Storage Attached Systems. I will give you a minute to think about his one.

测验答案 SAS 是什么? - Serial Attached SCSI 串联SCSI And, I hope you had some time to think of the correct answer for this one. And if you chose option (b), Serial Attached SCSI, you would be correct. SAS is the next evolution of SCSI technology.

What is SAS? 串联 SCSI (SAS) 是并口SCSI的终结者 提供了非凡的性能提升 新的工业标准 3-Gbps SAS x4 “宽” 端口提供累计高达12-Gbps的带宽 主机连接以及扩展 And as the successor to SCSI, SAS has a new roadmap with significant performance improvement milestones planned over the next several years as well as industry-wide acceptance behind it. 2007 is actually the year to be the tipping point where SCSI sales will switch over to SAS sales and you will see SAS taking the market lead over SCSI. And when it comes to the performance of SAS, it offers inherent performance enhancements with its four links within each connection. With these links, each SAS port currently at 3Gigabits can provide up to 12 gigabits per second total throughput. We will get into a bit more detail about this in the following slides.

理解SAS “宽” 端口 每SAS 线缆是3-Gbps “x4 宽” 在一个接口中包括4根线缆 1 2 3 每SAS 线缆是3-Gbps 到驱动器是独立线缆 “x4 宽” 在一个接口中包括4根线缆 2, 3 ,4 线缆仅在当前 I/Os 超出1号线缆通道性能情况下使用 Slide 11. Understanding wide SAS ports and how they work can be confusing so lets take a look at an image of a SAS port and see how data is actually transmitted and throughput is achieved. Lets first start from the drive perspective. And as you can see here the data is transmitted and received between the drives and the controllers in the system. And As I mentioned each SAS interface from the disk system to the server includes four links as you can see here - with links one, two, three and four. One question that comes up frequently is, "Will I always achieve sustained 12 gigabits per second throughput with SAS?" The answer is - sometimes.

理解SAS “宽” 端口 每SAS 线缆是3-Gbps “x4 宽” 在一个接口中包括4根线缆 1 2 3 每SAS 线缆是3-Gbps 到驱动器是独立线缆 “x4 宽” 在一个接口中包括4根线缆 2, 3 ,4 线缆仅在当前 I/Os 超出1号线缆通道性能情况下使用  What happens with these four links in data transmission is typical of an overflow process. For example, the first link will be transmitting data at 3 gigabits per second. If another block of data then needs to be written to disk for example, and link one is still busy - then link two will manage the overflow of data that cannot be transmitted by link one. So in this instance, you will see both link one and two being utilized.

理解SAS “宽” 端口 每SAS 线缆是3-Gbps “x4 宽” 在一个接口中包括4根线缆 到驱动器是独立线缆 “x4 宽” 在一个接口中包括4根线缆 2, 3 ,4 线缆仅在当前 I/Os 超出1号线缆通道性能情况下使用 4 1 2 3 And that's how it works. If link one is free, the first data block that needs to be transmitted will use link one - and will continue to use link one if that link remains free. However, if you had a busy time of the day where data is being written and read frequently, you would then likely see all four links transmitting data at the same time and the SAS port would achieve the full 12 gigabits per second throughput. In addition, SAS links are full duplex as you can see with link one having two paths to transmit and receive data, providing simultaneous data communication and doubling throughput as well.

SCSI 与SAS 适用于哪里? Server RAID sub-system SAS HBA or SCSI adapter Chip Controller SCSI/SAS Chip SCSI or SAS Controller SCSI/ SAS Chip So where do SCSI and SAS fit? Well, here is the big picture. In the local host itself on the left, I have my host bus adapter, or HBA, and this connects to my RAID array via a SCSI or SAS cable. With that and through the SCSI bus, we will send information to the target. Ethernet to Client workstations Dual Controller RAID with only one controller in use (B not used in this example). This RAID system has four SCSI buses with five drives on each bus.

Logical Unit Number 逻辑单元号(LUN) LUN是一种被SAS和SCSI支持的子地址编码 LUNs 是通过编码确认信息来选择 Controller Controller Host Adapter ID 2 ID 1 ID 7 There is an interesting concept to understand here and that is Logical Unit Numbers. LUNS are a type of sub-addressing. In effect it allows us to extend the SCSI addressing system. LUNS are selected through an identifier message. So in the example you see here, you may have an ID 7 which is your initiator, or host adapter, and it might send a message to controller ID 2. And in that message it may say, ok this message is for logical unit zero. In effect, Controller ID 2 becomes the initiator and it can initiate to the LUNs below it which really become the target. So by using LUNS, I can actually extend the quantity of device addresses, but not extend the number of ID numbers that are in use. Disk Drive Disk Drive Disk Drive Disk Drive LUN 0 LUN 1 LUN 0 LUN 1

小测验 LUN 是: Lower Unit Number 低单元编码 A type of sub-addressing 一种子地址编码 The SCSI initiation number SCSI初始号 A subnet mask 一个子网掩码 So now that we reviewed this concept, lets see if you remember what a LUN is. Is it (a) a lower unit number, (b) a type of sub-addressing, (c) the SCSI initiation number, or (d) a subnet mask.

测验答案 LUN是: - A type of sub-addressing 一种子地址编码 Well, you should have chose (b) a type of sub-addressing.

直连存储 通过SAS或SCSI连接到服务器 可以是 JBOD 或 RAID SCSI or SAS DAS Device Server Now you have an understanding of SCSI and SAS and that they can be used to directly attach storage devices to servers. Remember, a DAS device does not have to be external, but can be internal as well and will still use the SCSI bus. So know how about some terms I had talked about earlier? I had mentioned that storage could be a JBOD or a RAID – lets see what these terms are.

小测验 JBOD 是什么? Just buy old drives 只购买老磁盘 Java basic operating drive 基于JAVA的操作盘 Just boot every day 每日引导 Just a bunch of disks 一堆磁盘 Lets first get into the term JBOD and what this acronym stands for. Does JBOD stand for (a) Just buy old drives, (b) Java basic operating drive, (c) Just boot every day, or (d) Just a bunch of disks?

测验答案 JBOD是什么? - Just a bunch of disks 一堆磁盘 Well, your answer should be (d) Just a Bunch of Disks.

小测验 JBOD指的是什么? Redundant hard drives 冗余的磁盘 A virtual SCSI subsystem虚拟的SCSI子系统 SCSI disks that may be shared by multiple hosts 可以被多个主机所共享的SCSI磁盘 A group of hard drives in a common enclosure e)在同一磁盘柜内的一组磁盘 So you understand what JBOD stands for, but just as we did with SCSI and SAS – what does JBOD actually mean? What does JBOD actually refer to (a) Redundant hard drives, (b) A virtual SCSI subsystem, (c) SCSI disks that may be shared by multiple hosts, or maybe (d) A group of hard drives in a common enclosure?

测验答案 JBOD指的是什么? - A group of hard drives in a common enclosure 在同一磁盘柜内的一组磁盘 And the correct answer is (d) a group of hard drives in a common enclosure. As simple as the JBOD acronym is, the meaning of JBOD is quite simple and straightforward as well.

JBOD 在一个外部通用单元中的一组磁盘 不能提供容错 相对使用独立驱动器而言性能没有得到提升   JBOD 在一个外部通用单元中的一组磁盘 不能提供容错 相对使用独立驱动器而言性能没有得到提升  JBOD itself actually provides no fault tolerance. In fact, it provides not improvements in performance of basic drives. This is not to say that JBOD is not utilized – it is. Suppose, for example, you have a number of drives in different capacities. JBOD is a good way to utilize these types of drives.

  RAID 更好的选择; RAID But there is a better way and this better way is RAID.

小测验 RAID是什么? Redundant and independent disks冗余独立的磁盘 Real and imaginary disks真实可见的磁盘 Redundant Array of Independent Disks 独立磁盘冗余阵列 Remote abstract independent disks 远程抽象独立磁盘 So, what does RAID stand for? (a) Redundant and independent disks, (b) Real and imaginary disks, (c) Redundant Array of Independent Disks, or (d) Remote abstract independent disks ?

测验答案 RAID是什么? - Redundant Array of Independent Disks独立磁盘冗余阵列 Well, if you chose (c), Redundant Array of Independent Disks, you would be correct.

小测验 下面哪一个正确描述RAID? Redundant hard drives 冗余磁盘 A virtual SCSI subsystem一个虚拟SCSI子系统 Independent disks with data striped across them数据条带化分布的独立磁盘 A group of hard drives in a common enclosure 　在同一盘柜内的一组磁盘 Now that you know what it stands for, what does RAID actually provide – (a) Redundant hard drives, (b) A virtual SCSI subsystem, (c) Independent Disks with data striped across them, or (d) A group of hard drives in a common enclosure?

测验答案 下面哪一个正确描述RAID? - Independent disks with data striped across them 　数据条带化分布的独立磁盘 And your correct answer here is (c) Independent Disks with data striped across them.

RAID 条带化是RAID中的一个基本概念 多种级别 可以通过硬件或软件进行实施 是一种将多个驱动器划归到一个逻辑存储单元的组合方式 每种级别都具备不同的成本/性能/容错优势 可以通过硬件或软件进行实施 硬件对于整个系统的性能是最佳的方案 In RAID, striping is a fundamental concept and it is a method of grouping multiple drives into one logical storage unit so that these drives appear in fact as one single unit to the operating system. There are multiple levels or RAID and at each level there are specific cost, performance and fault tolerance benefits. The level of RAID itself is designated by a number. And its important to note that this numbering system does not indicate which RAID level is better but it really goes back to which RAID level will work best for your customers’ environment. RAID may be implemented by either software or hardware although hardware is best for overall system performance and reliability.

RAID Level 0 – 条带化 优点 高数据传输速率 低成本 缺点 当磁盘失效时无冗余 数据会丢失 Lets look closer at some of these RAID levels. RAID Level Zero is simply striping. It is simply taking a series of drives, grouping them so they are presented as a single device to the host and striping data across all these drives to improve performance. It certainly gives us a high data transfer rate and is relatively low cost. The disadvantage here is that there is no redundancy or high availability. This does not mean that this RAID level is not used. In fact, this RAID level is used with many applications in which performance, and not data redundancy, is important to your customer. 缺点 当磁盘失效时无冗余 数据会丢失

RAID Level 1 – 镜像 优点 可靠性 可用性 当出现错误时无需数据重建 缺点 额外磁盘所带来的高成本   RAID Level 1 – 镜像 优点 可靠性 可用性 当出现错误时无需数据重建 The next RAID level I want to talk about is RAID level one, or RAID mirroring. RAID level one offers excellent reliability and availability. And there is no rebuild of data necessary upon a failure. What it actually allows you to do is to have a mirror of your drives. However, your initial disadvantage may be the cost of additional drives as you have to double the amount of drives for the same amount of storage. 缺点 额外磁盘所带来的高成本

RAID Level 3 – 校验盘 优点 高数据传输速率 适用长的顺序数据 可靠性 可用性 缺点 对短数据不适合 对随机I/O不适合 RAID 3 gives us the concept of Parity. Parity is simply a way to ensure that the integrity of the data on the individual disk drives themselves. RAID 3 is very for long sequential records and has quite a good data transfer rate. It gives you the some reliability and availability because you have the parity drives. However, it is not good for short records and it is not good for random I/O. 缺点 对短数据不适合 对随机I/O不适合  

RAID Level 5 – 分布校验盘 优点 短数据记录或随机I/O 众多并发I/O 可靠性 可用性 缺点 不适用于大记录 So what is good for random I/Os and short records? And the answer is RAID 5. RAID 5 also uses the concept of parity, but instead of a single drive holding your parity, it is stripped across thedrive group. So this actually gives you a way to distribute your parity. As mentioned, RAID 5 is good for short record and random I/Os and it still gives you pretty good reliability and availability. It is not good, however, for long sequential records. 缺点 不适用于大记录

其他 RAID级别 RAID 1+0: implemented as striped subsystem whose segments are RAID 1 subsystems There is another RAID level and that is RAID 10. RAID 10 is implemented as a stripped subsystem whose segments are then RAID one. In fact, RAID 10 is taking the concept of RAID zero and RAID one and using them together. It certainly has its advantages. It has the fault tolerance of RAID one, but it approaches the performance of a RAID zero. And in a lot of environments, the initial disadvantage of the additional drive investment is offset by its performance and reliability advantages.

IBM RAID 总结 RAID 0: 最快和最有效的级别但不提供容错 RAID 1: 性能至上, 容错的环境下, 但需要2倍存储空间 RAID 3: Data intensive environments which access long sequential records, does not allow multiple I/O operations to be overlapped RAID 5: 对于多用户大量读取少量写入环境下的最佳选择. RAID 1+0: 高性能但需要2倍的磁盘以用于数据镜象 So let me summarize again. RAID zero is the fastest and most efficient level, but you do not have any fault tolerance. RAID 1 is for performance-critical fault-tolerant environments, but it does require that you double the amount of storage. RAID 3 is good for the data intensive environments which will access long sequential records, but because RAID 3 is lock-stepped together, it does not allow multiple I/O operations to overlap. RAID 5 is going to be your best choice for multi-user environments. And again RAID 5 is good for short records and random I/Os. And RAID 10 is great for performance, but it is important to note that you will have double the drive investment due to mirroring,

知识测验 对于一个主要是随机I/O并需要容错的应用你将选用哪种RAID 级别? RAID 1 RAID 3 RAID 5 So I have a question for you – of you were give this particular scenario in which you have an application that is mainly random I/O, but does require some fault tolerance, which RAID level would you choose?

测验答案 对于一个主要是随机I/O并需要容错的应用你将选用哪种RAID 级别? - RAID 5 And the correct answer here would be RAID 5 as it is good for random I/Os and does provide availability to your data.

RAID 直连存储 考虑 有限的距离 SCSI or SAS Server RAID So now I have talked a lot about the concepts of DAS, JBOD, of RAID and its levels, and the concepts of SCSI and SAS. Now lets look at some of the disadvantage of direct attach storage. One of the initial disadvantages is the fact that you have limited distance. And we know why this is – and that is dues to SCSI and SAS interface technologies.

RAID 直连存储 考虑 有限的距离 有限的设备数量(SCSI) 有限的冗余 SCSI or SAS Server RAID We also know that you have limited number of devices with a SCSI DAS configuration. And we also know that there is limited redundancy with DAS. So are there ways to overcome or improve this? And the answer is yes there is.

SCSI与SAS vs. 光纤通道 224 SCSI SAS Fibre Channel 线缆距离 3米 8米 最大10Km 双工 半 全 设备 12 ~ 16k (当带扩展时)) 224 -confirm 吞吐量 320 MB/s 3Gb/s* 4Gb/s 224 And the answer to this is something called fibre channel. The FC solution increases our distance significantly with cable lengths of up to 10 kilometers. In addition, the number of devices is increased with FC over SCSI. However, it is important to note, that with the introduction of SAS, the device limitations of SCSI are also removed. In lesson two, we will talk about a SAN environment where you can actually achieve the two to the 24 number of devices. Of course, I do not know of any implementation that have gone that high, but the potential is still there. And with FC, as well with the SAS interface and it wide ports, bandwidth is also dramatically improved. * Up to total 12Gb/s throughput with wide ports

知识测验 什么是 光纤通道? Transport mechanism for multiple protocols 多种协议的传输方式 　多种协议的传输方式 The FDDI specification 采用FDDI规范 Storage Area Network　存储区域网络 A network protocol　一个网络协议 Well, I have talked about FC. Now is the time for another knowledge check and here it is – What is FC? Is it (a) Transport mechanism for multiple protocols, (b) The FDDI specification, (c) Storage Area Network, or (d) A network protocol?

测验答案 什么是 光纤通道? Transport mechanism for multiple protocols多种协议的传输方式 - Well the correct answer is (a) – a transport mechanism for multiple protocols. And we will talk about this in more detail in the following slide.

什么是光纤通道? 一个适用多协议的传输方式 SCSI-3 与SAS Internet Protocol (IP) 其它 So what does that really mean? Well, it means that FC allows us to separate things such as your commands from the method that moves these commands. In fact, the FC transport mechanism allows you to move a variety of different protocols including the SCSI 3 and SAS specification command sets so your SCSI and SAS commands can move over a higher speed transport. This is equivalent to dial-up via DSL. I can still send the same information across both, but DSL is certainly a lot faster. The FC transport mechanism is similar to this type of concept. So what else can I move across it? I can moved the internet protocols across it, such as FICON, can be moved across the FC transport as well.

为什么使用光纤通道? Serial Transmission序列传输 高带宽 大写“B” 和 小写 "b" 更小的接口 消除并口传输带来的SKEW问题 高带宽 1, 2 和 4 gigabits per second (Gb/s) 大写“B” 和 小写 "b" Megabit 使用的是 小写 b (Mb) Megabyte使用的是大写B (MB) Well, why FC? One of the advantages is that it uses serial transmission. With serial transmission you have smaller connectors. You do not have to have 8, 16, or 32 or 64 pins to worry about. Serial transmission also eliminates the skew problems associated with parallel transmission that you may have. It is also important to note that in addition to FC, the SAS interface technology uses serial transmission technology as well. It provides high bandwidth with up to 4 gigabits per second. And while I am on the topic – 1 Gigabit is equal to 100 megabytes. You will typically see Megabit references with a capital M and lower case B and megabyte in all caps. When your are referencing megabytes, megabits or gigabytes and gigabits, make sure to look closely as, unfortunately, people tend to deviate from these.

为什么使用光纤通道? 可扩展性 非常多的设备 更远的距离 多种传输协议的方式 FC is scalable. We have already talked about that with FC, you can achieve up to two to the 24 devices. FC provides us with greater distances with cable lengths up to 10 kilometers and, of course, you can have multiple protocols.

为什么使用光纤通道? 采用交换 允许不同的速率 多种传输协议的方式 避免共享介质或带宽所带来的问题 自适应的4 Gb/s 允许向下兼容到 1Gb/s and 2Gb/s 多种传输协议的方式 FC also permits the concept of switching. And for many of you who are used to the idea of being able to use LANs or WANs, you are used to the idea of being able to use switches. It avoids the problems of shared bandwidth and mixed bandwidth. Some of your devices can be using copper wire, yes I said copper, as well as optical fibre And if you have noticed the spelling of fibre channel, you may have thought that fibre is mispelled. As FC is not necessarily always optical as it can also be copper, the unusual name for the spelling was derived. This is actually the French spelling for the word fibre. It is also – and again this is a key point – FC allows us to separate things such as your commands from the method that moves these commands

连接性 拓朴方式 设备数量 点对点 2 仲裁环 最多127 (in part 2) 交换网 最多16 million (in part 2) In this lesson we are going to talk about point-to-point topology where we basically have two devices, or in effect, a DAS environment. But there are two other topologies that you can use with the FC specification. There is first the arbitrated loop where you can have up to 127 devices, And we will talk about this in more detail in the 2nd training module. And the other one is a switched fabric which I will also discuss in more detail in part 2. Now lets look at point-to-point topology.

点对点 存储子系统 计算机 全双工 (FC example): 4Gbps + 4Gbps = 8Gbps 理论值) 100 MB/s 内存 N_Port0 N_Port1 100 MB/s 桥 TX RX RX TX 处理器 100 MB/s This may look very familiar to what you saw earlier when we were discussing the SCSI and SAS environments. In fact, in a lot of ways it is a similar type of concept. I replaced my SCSI or SAS controller with a host bus adapter, or an HBA. The HBA is capable of dealing with the FC specification. But consider this, on top of this, we are still using the same SCSI command set. So I have replaced the transport mechanism and I have increased the bandwidth, but yet I can still move from this point to the storage subsystem with SCSI commands. I/O Bus 全双工 (FC example): 4Gbps + 4Gbps = 8Gbps 理论值)

知识测验 光纤通道可传输 SCSI block data SCSI块数据 TCP 包 Parallel data packets 串口数据包 All network protocols　所有网络协议 And with that, lets get into another knowledge check. And I would like you to think about this one here. A fibre channel can transport (a) SCSI block data, (b) TCP packets, (c) Parallel data packets, or (d) All network protocols?

测验答案 光纤通道可传输 SCSI block data SCSI块数据 - And the correct answer here is (a) SCSI block data.

多模光纤 Input to fiber Output from fiber 150 meters In dealing with the fibre media, there are two kinds that you should be familiar with. There is multi-mode fibre. Multi-mode fibre has different wave lengths. Multi-mode 4Gb FC can go up to 150 meters depending on the wave length you are using. Input to fiber Output from fiber Multiple path-lengths, or modes, permitted by the fiber smear the shape of the pulse

单模光纤 Input to fiber Output from fiber And for longer distances, there is single mode fibre. Single mode fibre can go up to 10 kilometers. Input to fiber Output from fiber Single path-length, or mode, imposed by the fiber preserves the shape of the pulse

基本配置 Server RAID sub-system HBA SCSI Fibre Channel Chip Controller In this basic configuration with FC, we replaced our SCSI now with an HBA and using a FC cable going to the controllers as we had seen earlier. But notice that the controllers themselves are still capable of SCSI and that would be because, again, what we are transporting over this FC is the SCSI command set. In this case the controllers to the RAID devices are still talking and using SCSI. Ethernet to Client workstations Dual Controller RAID with only one controller in use (B not used in this example). This RAID system has four SCSI buses with five drives on each bus.

双控配置 Server RAID sub-system Sample SCSI/SAS subsystem HBA HBA Controller SCSI Chip HBA Fibre Channel HBA Fibre Channel SCSI Chip Controller Of course this implementation can be extended to dual independent controllers with automatic failover. In this case, each of these HBAs could be normally processing data or sending data to each of the controllers. And in the case of a failover, if one controller fails, the other controller now could handle the full I/O. Dual independent controllers with automatic fail-over for continuous availability in case one controller or one fiber link fails. Ethernet to Client workstations

Part I结束 Server 光纤 或SAS 子系统 – DS3000 and DS4000 FC/SAS HBA FC/SAS HBA Controller FC/SAS HBA FC/SAS And then finally what we see is the latest implementation and this would be with the IBM System Storage DS3000 and DS4000 systems where you see now that all the SCSI has been replaced with FC or SAS so we have FC or SAS transport to the controller FC or SAS from the controller to the RAID subsystem. Keep in mind though that we are still using the SCSI command set. Of course this implementation can be extended to dual independent controllers with automatic failover. In this case, each of these HBAs could be normally processing data or sending data to each of the controllers. And in the case of a failover, if one controller fails, the other controller now could handle the full I/O. RAID Controller Dual independent controllers with automatic fail-over for continuous availability in case one controller or one fiber link fails. Ethernet to Client workstations