Shulou(Shulou.com)06/02 Report--

A dazzling concept of technology at the moment

Storage devices have always been an important part of computer systems. For a long time in the past, users have been accustomed to using SAN storage as the only choice to save enterprise critical business data. The three-tier infrastructure based on SAN storage (server-storage network-storage device) has become the most commonly used infrastructure in the enterprise, also known as the traditional architecture. With the emergence of the Internet, the amount of information shows explosive growth, the speed at which human beings generate data exceeds the sum of all previous times, and the amount of data that needs to be stored is also growing rapidly. Big data, artificial intelligence, Internet of things and other new technologies have put forward new requirements for storage. So a series of new technical concepts and nouns about storage (including software-defined storage (SDS), superfusion (HCI), distributed storage, Server SAN, etc.) have emerged in the market, which suddenly burst into the eyes of users. In the face of some specious definitions and the hype of various manufacturers, I believe you would like to know what is the essence behind these concepts? Can you solve those problems? And is there a suitable application scenario?

Based on the above reasons, this article will introduce several concepts mentioned above to help you better understand these technologies or concepts and make your own choices.

The concept explains what software-defined storage (SDS) is

The full name of SDS is Software Defined Storage, which literally means software-defined storage. The definition of SDS can refer to the Global Network Storage Industry Association (Storage Networking Industry Association,SNIA). SINA formally listed Software-defined Storage (SDS) as the object of study in 2013.

SINA defines software-defined storage (SDS) as a virtualized storage with a service management interface. SDS includes the function of storage pooling and the data service characteristics of the storage pool can be defined through the service management interface. In addition, SINA also proposed that software-defined storage (SDS) should have the following characteristics:

High degree of automation-reduces the operation and maintenance overhead of the storage infrastructure by simplifying management-standard interfaces that support API management, publishing, and operation and maintenance of storage devices and services virtualize data paths-supports a variety of standard protocols and allows applications to write data scalability through block, file, or object storage interfaces-storage architecture has the ability to scale seamlessly The scaling process does not affect availability and does not lead to performance degradation transparency-storage should provide users with available resources and overhead to manage and monitor storage what is distributed storage

There is actually no clear definition of distributed storage, or even a unified name, which in most cases is called Distributed Data Store or Distributed Storage System.

The definition of Distributed data store in Wikipedia is: distributed storage is a computer network, which usually stores information in multiple nodes by way of data replication.

The definition given in Baidu encyclopedia is: distributed storage system, which stores data on multiple independent devices. The distributed network storage system adopts a scalable system structure, uses multiple storage servers to share the storage load, and uses the location server to locate and store information. it not only improves the reliability, availability and access efficiency of the system, but also is easy to expand.

Although the definition of distributed storage is not exactly the same, one point is unified, that is, distributed storage distributes data in multiple nodes, and nodes provide storage services through network interconnection. This is clearly different from the way traditional centralized storage places data centrally.

What is Server SAN?

The concept of Server SAN was first put forward by WIKIBON (a famous storage consulting community abroad). WIKIBON defines Server SAN as the combination of computing and pooled storage resources, and multiple storage media are directly mounted to multiple independent servers. These directly connected storage devices (DAS) are interconnected through high-speed networks (InfiniBand or low-latency Ethernet) and ensure storage consistency through software. Server SAN can use mixed storage media, including mechanical hard drives and SSD hard drives. At present, the mainstream perception of the market is to use Server SAN as the architecture of SAN, but storage no longer uses dedicated hardware and controller architecture. However, in the ideal situation defined by Wikibon, enterprise applications run on Server SAN and can achieve high availability, and even classify super-converged products as Server SAN, but because this concept is not common in the market, it does not focus on introduction.

What is superfusion?

Refer to the definition of superfusion in Wikipedia:

Super converged Infrastructure (hyper-converged infrastructure) is a software-defined IT infrastructure that virtualizes all elements of a common "hardware-defined" system. The smallest set of HCI is: virtualized Computing (hypervisor), Virtual Storage (SDS), and Virtual Network. HCI typically runs on standard commercial servers.

The biggest difference between Super converged Infrastructure (hyper-converged infrastructure) and converged Infrastructure (converged infrastructure) is that in HCI, both storage underlying abstraction and storage network are implemented at the software level (or through the hypervisor level), rather than based on physical hardware. Because all software-defined elements revolve around hypervisor implementations, all instances on a super-converged infrastructure can jointly share all managed resources.

Difference and relation

Software-defined storage (SDS) focuses on storage resource virtualization and software definition. First of all, software-defined storage (SDS) is different from traditional "hardware-defined" storage in shape. It does not rely on exclusive hardware and allows storage software to run on a general-purpose server, avoiding hardware binding and effectively reducing hardware procurement costs; it has standard API interfaces and automation tools to effectively reduce the difficulty of operation and maintenance. Storage resources virtualization, support a variety of storage protocols, can integrate enterprise storage resources and improve the utilization of storage resources. But in terms of definition, software-defined storage (SDS) is not necessarily distributed or centralized in terms of deployment form, that is to say, SDS is not necessarily distributed storage (although the common software-defined storage (SDS) is more distributed), SDS storage may be run on a stand-alone machine and do not store data separately through the network. The scalability of this form of software-defined storage (SDS) may have great limitations.

Distributed storage, its biggest feature is multi-node deployment, data is distributed through the network. Distributed storage is characterized by strong scalability, through multi-node load balancing to improve the reliability and availability of the storage system. In contrast to software-defined storage (SDS), distributed storage is not necessarily software-defined, but may be hardware-bound, such as IBM XIV storage, which is essentially distributed storage, but is actually delivered through dedicated hardware. Then there is still the problem of hardware binding and high cost of ownership.

From the definition of Server SAN, the definition of storage part seems to be the definition of distributed storage, but Server SAN emphasizes the resource combination of computing and storage, that is to say, unlike software-defined storage (SDS) and distributed storage, the later definition already has the embryonic form of super-fusion concept, but this concept has not become the mainstream.

The super-converged infrastructure is defined to include software-defined storage (SDS), which has the ability of hardware decoupling and can run on general-purpose servers. The super-converged infrastructure is similar to the concept advocated by Server SAN, computing and storage convergence, through a fully distributed architecture, effectively improve system reliability and availability, and have the characteristics of easy scalability.

SMTX OS product architecture

As many readers are still confused about the composition of superfusion, the following is an example of SmartX's hyperfusion software SMTX OS to illustrate the relationship between distributed storage and other modules.

Among them, distributed block storage, SMTX ZBS is the core component of SMTX OS hyperfusion software. It uses a fully distributed architecture and is fully in line with the concept of software definition.

SMTX ZBS distributed storage architecture

In addition, the super-converged infrastructure has been further extended, which emphasizes virtualized computing (hypervisor) as the core and software-defined integration including virtualized computing, software-defined storage and virtual network resources. From the author's point of view, the super-converged infrastructure has more possibilities in the future, which can promote the convergence of computing, storage, network, security, disaster recovery and other IT services, reduce the complexity of IT infrastructure, and reshape the "software-defined data center".

The technical core of a super-converged infrastructure

Super-converged infrastructure can be said to integrate many excellent features of software-defined storage (SDS), distributed storage, and even Server SAN, and it is not difficult to see that a keyword is indispensable in these concepts-storage. Obviously, the technical core of super-converged infrastructure is storage, which carries user data, and its importance is self-evident. In addition, various advanced functions of super-converged (HA, online migration, etc.) are extremely dependent on storage capacity, and the scalability of storage directly limits the scale of super-converged products. The manageability of storage also restricts the flexibility of the platform. Considering that in the super-fusion form, computing and storage run in the server with the same software stack, and there is a resource competition between them, under the premise of ensuring the stability and performance of storage software, the less overhead the better.

Finally, we summarize the storage requirements of the super-converged infrastructure, which can be used as some references for examining the storage capacity in the super-converged infrastructure:

Software definition-unbinding hardware, can expand richer functions through upgrade, high automation capability, fully distributed architecture-good scalability, eliminate single point of failure risk, high reliability-intelligent fault recovery function, rich data protection means, high performance-support for multiple storage media, make full use of the performance of new hardware highly integrated-simple architecture and easy to manage

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.