Abstract: In this article, we will take a look at readers carefully analyze the current driving force of the market, what are software-defined storage; software-defined storage What are the different types; and each of the respective advantages of different types of shortcomings.
In the current storage areas, I am afraid Nothing 'Software-defined storage' (SDS) The concept of the term is more cause more disagreement the view of storage software on industry-defined yet have a generally accepted definition , SDS storage vendors are specific to storage vendors to build the SDS this definition to match their storage products. as a result, every storage vendor seems to offer different SDS products.
The closest software-defined storage market has been consistent for the SDS formed a consensus that: it is more conceptual market structure (marketecture), rather than a simple architecture.
Software-defined storage hardware store data from its implementation separation management software storage software itself is independent of the hardware. Typically, the control panel is stored (but not always) separated from the data panel.
This broad definition of each type of change to promote the current store. So, this would require software-defined storage market business consumers themselves to determine which product is best for them.
The driving force behind the development trend of SDS
All storage systems have been changed by software-defined merely: This software has become portable.
Storage-system software has always been tied to its management hardware. When the hardware capacity or performance overload, we must implement a replacement, then the client organization will need to replace hardware together again buy software licenses.
Worse, the storage system architecture to create isolated islands of silos. Unique infrastructure makes everything ranging from storage configuration, data protection, disaster recovery, technology updates, data migration, including power and cooling are becoming increasingly untenable. and coupled with the rapid growth of data trends, and storage requirements for the growing amount of data growth, the existing architecture makes storage management system is too complex, difficult and expensive costs, and ultimately difficult to maintain .
Several technical factors also promoting software-defined storage market has become a phenomenon made its contribution. First, is the direct result of x86 computing architecture to enhance performance continues to improve and caused .x86 core performance for specific storage functions of availability, leading to the x86 architecture standardized storage system.
Help promote the SDS Another factor is server technology, desktop desktops, applications, and networks (SDN) for x86 virtualization generally accepted that it has helped IT Adaptations receiving image data from the hardware it resides in the separation.
The popularity of cloud technology for the promotion of software-defined storage market also plays an important influence. According to industry standards and commercialization of hardware, cloud data center needs a new, low-cost storage infrastructure.
Other technical factors driving the SDS include: server-side flash technology, as well as memory and allows the server to store and other physical server hosts for transparent sharing software.
All of these technical factors changed the distinction between server and storage hardware, while accelerating storage software portability and flexibility, and last but not least: also reduces storage costs radically.
SDS pros and cons of different categories of analysis
Given the current industry for software-defined storage has not yet formed a generally accepted standardized definitions, making a variety of related technologies have been defined in the software storage market rise. In our opinion, the current SDS on the market can be divided into the following four class:
· Hypervisor based on the SDS
* Ultra converged infrastructure (HCI) SDS
· Scale (Scale-out) objects or documents S DS
In these types of products, that there are significant differences, there also has significant similarities and there are several products that may be suitable for a variety of classification category; while some products may be the only product of its category category, such as PernixData or Saratoga Speed.
Since SDS to focus on flexibility, simplicity, scalability, performance and total cost of ownership (TCO), we will use these criteria to evaluate each of the different advantages and disadvantages of each method of SDS.
Based on the SDS Hypervisor
VMware invented this type of SDS products through its VMware vSphere Virtual SAN enterprise storage solutions. This is the only category of this particular product category Virtual SAN architecture is designed to be part of vSphere as a function of running vSphere and is compatible with all vSphere virtual machines and virtual desktops desktops virtual SAN running ESXi layer, which means that it is not a virtual storage device and does not require a virtual machine to execute.
Hypervisor-based advantages of the SDS:
Flexibility virtual SAN with a hard disk drive (HDD) and solid state drives (SSD) including DIMM-based flash drives, PCIe, SAS, SATA or NVMe compatible work .VMware virtual SAN in mixed mode support HDD and SSD; or full flash All SSD solid state hard drive model.
Scalability and performance. Virtual SAN highly scalable, while providing high levels of performance. Its extensible through vSphere clusters, each cluster server cluster can support up to 64 groups of vSphere hosts. Each host vSphere supports about 140 TB raw storage capacity, each cluster server cluster with 8 PB of raw storage capacity. in terms of performance, each virtual SAN host can provide up to 90,000 IOPS, each cluster can produce more than 5 million of IOPS.
Simplicity Virtual SAN is simple, because it integrates natively integrate itself became part of VMware stack for vSphere administrators, this feels like all the other functions, very intuitive. In each virtual machine based on the automatic virtual SAN storage tasks, such as provisioning, snapshots / data protection, high availability, stretch clusters, disaster recovery and business continuity even if the data is migrated to a virtual SAN can be relatively easily accomplished through vSphere storage VMotion .
Total Cost of Ownership (TCO). Compared with traditional storage architecture that TCO should be less.
In a storage system, compared to the same storage drive in a server, the cost savings from the price differences main storage drive (HDD and SSD) In the storage system, this type of drive prices generally three times more expensive virtual SAN some other cost advantages from foreseeable prorated, unified storage management, unified data protection, disaster recovery and business continuity; and consolidated storage networks.
Based on the SDS Hypervisor disadvantages:
Flexibility problems. Virtual SAN is a closed loop of the SDS, which can only be compatible with VMware vSphere or later version 5.5. ESXi deployment earlier, other hypervisors hypervisor or physical machines are not compatible with Virtual SAN . it can not be virtual or physical machines that are not part vSphere cluster use. it has a DIY (DIY) virtual SAN elements. for example, in low-cost commodity hardware to run will be limited to the VMware hardware compatibility list (HCL). If the hardware is not in the list, which is not supported.
Scalability and performance issues. Virtual SAN clusters no more than 8.8 PB. If you need more capacity, which is not a good choice. If a VM needs IOPS than IOPS available on the vSphere host its 90,000 more more, which can be obtained from other nodes in the cluster, but in terms of delay or will be a considerable impact cross clustered storage performance is another problem. most virtual SAN cluster using 10 Gbps to 40 Gbps Ethernet and TCP / IP host and interconnection. the architecture is essentially a deterministic system bus to replace a non-deterministic TCP / IP network, so the network delay between hosts become highly variable, unless the use of more advanced cluster and faster interconnect, otherwise from a cluster host to another storage performance will be highly variable and inconsistent.
Some things are not so simple. From isolated storage environment into a purely virtual SAN, you first need to be a non-VM image to a virtual machine. For non-vSphere environment this would be a fairly time-consuming process.
TCO issues. Until the recent release of version 6.2, virtual SAN has been the lack of data deduplication and compression. This compared with those data reduction capabilities of SDS products, increases the cost per usable TB Further in a particular vSphere host cluster to ensure data and VMDK other clusters remain available in case the host fails to meet the current needs of multiple copies of the image. best practices require at least two copies of the original data, but many administrators Select three copies. this approach eliminates the price advantage drive. and because virtual SAN is unique options vSphere, the license fee will rise sharply.
Super converged infrastructure (HCI) SDS
HCI combines servers, storage, networking and hypervisor, and packaged them to the cluster nodes. HCI is designed to eliminate it-yourself integration trouble expensive server hardware, the need for storage over-provisioning, high availability . the problem, complex storage management and hardware compatibility problems many HCI options from the server and other vendors, including: Atlantis, Cisco, Dell, EMC, Fujitsu, Gridstore, Hitachi, HPE, IBM, Lenovo, Maxta, NEC , Newisys, Nutanix, Quanta, Saratoga Speed, Scale Computing, SimpliVity, StarWind, StorMagic and SuperMicro.
Super Fusion Advantages Infrastructure (HCI) SDS is:
Flexibility With a virtual SAN VMware, a virtual machine can control the storage administrator fact, some HCI is based on the deployment of VMware vSphere and Virtual SAN, including VMware's EVO:.. RAIL reference design has several HCI suppliers the hypervisor hypervisors options, including some 'bare metal' from vSphere, Hyper-V, KVM or XenServer to be used for Linux use Docker container or application virtualization (Saratoga Speed) of the product, without having to manage the program. HCI deployment allows for many different capacities node within the cluster size. there are several pure software products, such as Maxta, StarWind and StorMagic.Maxta and including Dell, Quanta and SuperMicro, including most of the major server vendors reached Partnerships.
Scalability and performance .HCI extended zooming is just as easy to add nodes to the cluster. Zoom just need to add storage capacity drives (HDD or SSD) to a maximum value of a node or add additional nodes. Each product has HCI However, most products are well extended to the PB level, and with each server node linearly add performance to the cluster; the limitations of its own scalability and performance.
Simplicity. Insert, open, configuration, and then you're done. Few systems are so simple, no DIY.
Total Cost of Ownership (TCO). Similar to the virtual SAN VMware. HCI many vendors, including duplicate data delete and inline compression, depending on the data, you can reduce the total capacity requirements up to 83% to 90%, thus greatly reducing the TCO .
Super converged infrastructure (HCI) SDS drawbacks:
Flexibility is a closed loop problem .HCI SDS system, so they are compatible only with the storage server nodes in the cluster to any physical or virtual host cluster do not belong to HCI HCI will not be accessible storage (of course there are exceptions:.. Saratoga Speed offers their so-called all-flash super converged infrastructure (UCI), which may be outside the cluster as a target UCI physical or virtual host.)
Cluster hardware is limited to HCI provided by the supplier or only HCI software certified hardware. And VMware virtual SAN, as there will be vendor lock-in, and changing suppliers need everything to migrate from the old HCl to the new, which It will undoubtedly be a very time-consuming and tedious process.
Scalability and performance issues. HCI limit capacity is supported by the cluster nodes and the number of nodes supported by the capacity of each amount. If a VM in vSphere hosts its IOPS required exceeds its available 90,000 IOPS, can be obtained from other nodes IOPS, but in terms of delay or will be a considerable impact cross clustered storage performance is another problem. most virtual SAN cluster using 10 Gbps to 40 Gbps Ethernet and TCP / IP interconnection with the host, so the network delay between hosts become highly variable.
Some things are not so simple from an isolated storage environment into a cluster HCI, first non-VM images and VM virtual machine into HCI, or Docker container vessel. This will be a fairly time-consuming process.
TCO issues. Virtual SAN, like, HCI ensure that data, virtual machine images, VD images and Docker container container container images on a specific HCI node, there is still a problem the rest of the cluster available aspect is. Now, this requires multiple copies of the image. best practices require at least two copies of the original data, but more common is the three copies, thereby increasing the overall capacity requirements and associated costs.
Storage virtualization SDS
On software-defined storage market, storage virtualization SDS is arguably the most mature of the SDS. Since the early 2000s it has appeared, then called storage virtualization storage virtualization is mainly SDS entire storage software stack, including all storage services, optimization runs on the x86 architecture and will host converted to a powerful full-featured memory controller to store their virtual servers and external storage systems to create a different capacity, data protection strategy and performance characteristics . or more virtual storage pool storage virtualization nature SDS is to x86 server into the storage system; some of the products also can be used as a virtual machine's virtual storage devices (VSA) SDS running storage virtualization is mainly Scale-up (longitudinal direction. extension) architecture, but some products Scale-out (scale) is also very good. their high cost of dedicated hardware architecture for the elimination and take advantage of server drive lower costs, re-use the old storage system, and simplify data migration some of the more well-known suppliers / products include: DataCore Software SANSymphony, EMC VIPR, IBM SVC, Microsoft Windows 2012 R2 (and later versions), NetApp Ontap Cloud, Nexenta Systems NexentaStor, QuantaStor and Starwind Software.
SDS advantage of storage virtualization:
Flexibility which is suitable for most x86 physical host or virtual machine, as long as the hardware or hypervisor has been certified, supported by the vendor. Convert all storage located behind it to a virtual storage pool, making the old storage re-use lateral extension version allows access to any physical or virtual machine node. multi-copy image is not necessary, in order to prevent a single controller failure, although it is available. similar HCl, storage virtualization can be used as SDS software or hardware bundled with the server.
Scalability and performance are multidimensional scaling, cluster, each node can be enlarged, and that more nodes can be added to the extended zoom. Under normal circumstances, the storage virtualization SDS corresponds to the most active silo storage system .
Simplicity. When bundled with hardware, storage virtualization SDS is a very simple storage system that utilizes commercially-shelf hardware, better scalability, and in some cases, while providing block (SAN) and file (NAS). but in the end, it is still at a low-cost container vessel isolated storage systems.
Total Cost of Ownership (TCO). In the storage virtualization SDS, the biggest cost savings come from commodity hardware and drive-based servers. Another cost savings from inline data reduction technology. Compared to peer storage systems, large most storage virtualization SDS will produce a more favorable TCO.
Storage virtualization SDS disadvantages:
Flexibility problems. Most storage virtualization SDS can only be run by the supplier qualification and support the commercialization of specific hardware can be run as a VSA product requires hypervisors certification and require vendor support.
Scalability and performance issues. On paper, these systems support the massive scalability of capacity, but a pragmatic approach is a bit different storage virtualization capacity by SDS X86 server limitations. In the performance drops to an acceptable below the level before, each server can handle only low capacity ability scale storage virtualization SDS also be clustered limited because the number of supported storage controller node is limited. performance may also be subject to the same restrictions constraint.
Some things are not so simple. Storage virtualization SDS is mainly DIY system integration testing integration needs, the need to ensure quality assurance software and hardware compatible with normal work may be required to implement the deployment of professional services or system integrators.
TCO issues. Depending on the vendor, its licensing may be a bit expensive. Not all storage virtualization SDS products offer inline deduplication and compression capabilities. These problems will undoubtedly TCO harmful effects.
Scale (Scale-out) objects or documents S DS
Recently, we have seen the introduction of the S DS scale objects. The objects are stored as objects data management, including data, metadata, and a unique identifier. There are quite a number of object storage vendors, as well as OpenStack Swift and Ceph two on the market paragraph open source products.
Scale file SDS is a highly scalable NAS, often with special features such as elasticity or unique metadata object store (Qumulo), etc. Some lateral extension file SDS products are actually stored in the object (Exablox) of on the other, in essence, it is IBM's General parallel file system (Spectrum Storage) cluster scale-out deployment.
Scale (Scale-out) objects or documents S DS advantages:
All scale SDS architecture flexibility is designed for x86-based servers. Some products may be certified by the suppliers of hardware as software deployment, and some other products can be bundled with the server hardware. They are not designed to VSA, and generally intended for secondary or non-mission critical applications.
Many object or file SDS scale product deployment for Hadoop HDFS as storage. By reducing the required number of copies of the mirror and allows the re-use of NFS or SMB data, in order to significantly reduce the HDFS storage costs.
Scalability and performance are multidimensional scaling: each node can be individually scaled, generally cluster node itself can add capacity or performance will never be close to the performance of high-performance block storage.
Simplicity. When bundled with hardware, scale object or file stored settings, configuration, and management is very simple. As the deployment of software to realize the need for manual DIY system integration. These two types of commercially available hardware, with special scalability - in the case of scale object storage through data erasure coding with unmatched resilience and longevity.
Total Cost of Ownership (TCO). Both types of designs are low cost and provide very little additional features, licensing is usually based on annual purchase. Scale object storage can be reduced by erasing codes per GB the overall cost than traditional RAID because of data protection and replication requires less overhead.
Scale object or document S DS disadvantages:
Flexibility problem, whether it is delivered as a software or hardware bundled with the hardware vendor must obtain certification and support.
Scalability and performance issues. Scale SDS files generally do not have the same high-scale object storage expansion of the scale, but the object will be slightly delayed. Object store already has a significant additional delay from the metadata and data resiliency features. both types are best suited for high performance does not require auxiliary applications.
Some things are not so simple. When the scale object storage SDS file or to be purchased as software, which is a DIY project, it may require special skills, professional services or system integrators.
In addition, when these types of SDS is used to support the application, such as when archiving, data must be migrated from its current location; some vendors have a product can do this, but most products need to rely on third-party software.
. TCO issues of data reduction - deduplication or compression feature is currently not available in the scale of objects SDS products, small scale file SDS have this feature which increases the total cost of ownership.
Software-defined storage market summary
SDS is a broad marketing term, currently has a variety of software-defined storage market, each of which each has its own advantages and disadvantages.
Organizations want to select the appropriate SDS to meet their operational needs, the application must, storage capacity and performance requirements, skill requirements, and how software-defined storage market can deal with specific issues which have an accurate understanding.
www.get-top-news.com Get Top News, Latest News and Current News from get-top-news.com. Latest Current News: U.S., China, World, Entertainment, Health, Finance, Business, echnology, Travel, Politics, Sports.Traditional Chinese: 定義軟件定義的存儲市場的力量