IBM cloud storage technology

In September, IBM announced the launch of the latest storage Strategic Plan in the cloud computing field. This application, known as "enterprise-level smart cloud storage", is a private cloud-based storage and archiving technology, designed to provide applications for enterprise customersProgram. This type of cloud storage is mainly achieved through storage virtualization. This is different from the existing low-cost cloud storage application environment provided by EMC and other vendors to customers. It stores two or three-level data copies or applications in development, utilization, and test environments. IBM cloud storage is an integrated solution based on IBM Storage Virtualization combined with a variety of storage devices.

The IBM cloud computing solution is an advanced infrastructure management platform developed by the IBM cloud computing center after years of exploration and practice. This solution combines the latest technologies in the industry to fully reflectCloud computingConcept, has been successfully running within IBM for many years, and has many customer cases around the world.

This solution can integrate the existing infrastructure of enterprises and build their own cloud computing centers through virtualization and automation technologies, it enables unified management, distribution, deployment, monitoring, and backup of Enterprise hardware and software resources to break the exclusive resource of applications and help enterprises realize the concept of cloud computing.

The IBM cloud computing solution consists of the following parts:

Hardware and software resources that need to be included in the cloud computing center. The hardware can include x86 or power machines, storage servers, switches, routers, and other network devices. The software can include various operating systems, middleware, databases and applications, such as Aix, Linux, DB2, websphere, Lotus, and rational.

Cloud computing management software and IBM Tivoli Management Software. Cloud computing management software is developed by the IBM cloud computing center to provide cloud computing services.

Cloud computing consulting service, deployment service, and Custom Service. The cloud computing solution can be developed based on the customer's specific needs and application scenarios, so that the cloud computing management platform can be integrated with the customer's existing software and hardware.

Is the architecture of IBM cloud computing system

Cloud computing solutions can bring the following benefits to enterprises:

Unified management of IT infrastructure (servers, networks, storage, software, etc.) to effectively control IT resources

Integrate standard processes and templates into IT management to reduce O & M risks

Share resources to improve resource utilization

Reduce power consumption

Reduce system maintenance costs

Quick Response to enterprises' demands for IT resources

We can see that the storage virtualization of IBM isCloud computingPart of the solution.

So what is Storage Virtualization?

Add a Virtual Layer Between the physical storage system and the server to manage and control all the storage and provide storage services to the server.

Servers do not directly deal with storage hardware. The increase, decrease, change, split, and merge of storage hardware are completely transparent to the server layer.

Hiding complexity

Allows you to integrate and use existing functions.

Get rid of the limitations of physical capacity

Storage Virtualization of layer-3 Infrastructure

Modern storage virtualization technology pools the products of heterogeneous storage vendors in a special way, so as to provide advanced functions such as data migration without downtime and simplified configuration.

This level of abstraction can be implemented in the layer-3 Infrastructure of servers, storage networks, and storage controllers.

Server-based

Some of the earliest Storage Virtualization did not appear in the storage infrastructure, but inside the server. More specifically, it appeared in the operating system of the server.

Although the file system helps to call information easily, as more and more file systems are established, the storage space of the physical Lun will be used up and another Lun needs to be created, and specify it to the operating system to continue storing files. To learn which data is stored on which Lun, the operating system needs to assign a volume number, name, or identifier to each Lun. In Microsoft Windows, what you are most familiar with is the letters assigned to each volume, such as c: \ or D: \. in UNIX, these letters are/dev/HDO or/dev/hd1. As applications and users create more and more files, more and more volumes are required, and management will become very difficult soon.

Therefore, the operating system vendor proposed the concept of logical volume manager (LVM. Similar to how the file system combines blocks to provide files, LVM combines volumes or Luns to provide a larger and more flexible storage pool for applications, as shown in figure 1. When the space of the LVM is close to the light, you can concatenate or add another volume to expand the space without re-configuring or disabling the application. On the contrary, if you want to split a large volume into smaller blocks, LVM can help you partition to distinguish different information. For example, a block is used for the operating system, another block is used for user data.

Server-based storage virtualization was initially integrated into the operating system as software and is still very popular today. The following are the main advantages of this method:

Because of the built-in system software, server-based storage virtualization is highly configurable and flexible.

Because most operating systems include this feature in their own system software, it is very cheap.

The storage infrastructure does not need to be configured with other hardware and can be used with any device identified by the operating system.

Server-based virtualization also has many drawbacks:

Although it helps to maximize the efficiency and recovery level of storage resources, it can only be optimized based on servers.

The image, splitting, and computing parity tasks need to be processed separately, occupying valuable CPU resources and

Each operating system manages file systems and volumes in different ways. Enterprises that use multiple it supplier products must have different technologies and processes, increasing costs.

During data migration or replication (both local and remote), it is very difficult to ensure data tracking and protection throughout the environment.

Most operating system vendors, such as Microsoft, IBM, HP, and RedHat (Linux), have at least certain capabilities to provide virtualized storage resources. Symantec and other vendors provide more advanced, server-based storageVirtualization.

Storage Network-based

In the late 1990s S, the emergence of network attached storage (NAS) and storage area network (SAN) allowed the separation of disks (and their controllers) from servers, all applications in the IT environment can share storage resources more effectively. Storage networks have become the hub for servers and storage devices to exchange all information. Some storage vendors believe that this is the best place to manage virtualization.

Network-based storage virtualization is embedded with intelligent management of storage resources at the network layer. The actual storage resources between the abstract server and the storage array can be either in-band or out-of-band.

The in-band mode, sometimes called Symmetric, embeds the virtualization function into the I/O (input/output) path between the server and the storage array, as shown in 2, and can be configured in the San switch itself or a dedicated device. All I/O (input/output) requests are transmitted through the device together with the data. The server interacts with the virtual device instead of the storage device. The virtual device can analyze the request, view its ing table, and execute storage device I/O in sequence. These devices can not only convert Storage requests, but also use their on-board memory to cache data, provide data usage metrics, manage Replication Services, coordinate data migration, and perform simplified configuration.

The out-of-band mode, which is sometimes called asymmetric, is not strictly limited to the I/O (input/output) path as the in-band mode. It performs a specific search with a specific virtual SAN Switch, 3. The server maintains direct interaction with the storage layer through a smart switch. The out-of-band device maintains the San ing (usually called metadata) of all San-connected storage resources and instructs the server where to find them. In this two-step process, the server uses specialized software or proxies because commands must be sent through San to work. Because data does not pass through virtual devices, the performance is only slightly affected. However, data caching and other functions cannot be used.

Both the in-band and out-of-band modes can provide storage virtualization with the following features:

Pools storage products of heterogeneous suppliers into a seamless access pool.

Perform replication between devices that are not similar.

Provides a single management interface.

However, only the in-band method can cache data to improve performance.

Figure 3: out-of-band network storage virtualization uses smart switches to maintain interaction between servers and storage arrays.

Vendors such as IBM, EMC, and falconstor provide network-based storage virtualization solutions.

Basic storage controller

It is applicable to enterprise-level storage arrays of large enterprises in terms of performance and functions. It has the virtualization function to enhance physical storage resources (to a certain extent, it is better than other systems ). For example, raid can protect data and avoid disk faults. The Host Storage domain can virtualize the frontend connection port so that multiple operating systems can use the same physical port. Many enterprise-level devices adopt an advanced switching architecture that supports multiple physical connections to disk drives, provides balanced performance and elasticity, and supports logical segmentation of the internal global cache, ensure service quality and security.

Over the past 15 years, various forms of virtualization technology have been applied to storage controllers. Some storage array vendors believe that these features should be extended to storage resources outside the device. Controller-based storage virtualization came into being (see figure 4 ).

This new generation storage array allows other heterogeneous vendor storage arrays to connect directly to their own controllers. As a result, external storage resources on the basis of the "appearance" can be managed in the same way as internal disks. This method has many advantages, including the absence of re- ing of Luns or ranges, which means that a layer of management is reduced and network complexity is greatly reduced. In this way, the first-in microcode software installed on the storage controller becomes an external storage resource, just like inside the array, and the host does not know their actual connection location.

Figure 4: controller-based storage virtualization makes external storage the same as internal storage.

Hitachi Data Systems (HDS), netapp V series, Sun Microsystems, and HP all adopt this form of storage virtualization.

Virtualization architecture selection-superior to inferior

The virtualization architectures are in sequence: in-band, out-of-band, and storage controller-based.

SVC-advantages of in-band block-level virtualization

IBM San volume controller for Storage Virtualization

IBM San volume controller (SVC) is an integrated software and hardware product, including IBM System X servers, Linux kernel-based storage operating systems, and professional virtual storage software. SVC supports more than 120 disk storage systems, including IBM and non-IBM. By integrating storage capacity into a storage resource pool, you can simplify your storage infrastructure, manage the lifecycle of information and maintain business continuity.

SVC logical architecture

SVC superior performance:

The fastest controller in the SPC Benchmark Test

SVC has the fastest SPC-1 benchmark value (SVC 4.2: 272,500 iops)

SVC has the fastest SPC-2 benchmark value (svc4.2: 7080 Mb/s)

Many cases show that SVC performance is significantly improved (up to 10 times)

SVC can improve the access performance of the storage part in the Application System

Especially for old storage devices with small caches

For cache-friendly business flows, performance improvement is the most obvious

The performance is equivalent to that before use for business flows that are not cache-friendly

The performance growth of SVC shows a linear trend, for both iops and bandwidth.

Introducing SVC to a storage network usually does not negatively affect performance.

In the worst case, performance is reduced by about 0.6%.