Shulou(Shulou.com)06/03 Report--

Note: the content of this article is sorted out from the speech of SmartX CTO Zhang Kai at the launch of SMTX OS 3.5 new products.

We have recently released a very important product version-SMTX OS 3.5, which is introduced in detail here. Before that, let's briefly introduce the situation of super-integrated external market and the market progress of SmartX, which are the important basis for us to do product iteration and market expansion.

It is inevitable that the superfusion will subvert the traditional IT architecture.

If we say that super-integration is still in the stage of market education in the past few years, with the increase of customer cases in the past two years, the applicability and value of users to super-integration have become more and more clear, and we can clearly feel that the market has entered a stage of real explosive growth.

Both Gartner and IDC predict that the global super-integration market will maintain a rapid growth rate of nearly 50% compound growth rate and reach a market size of nearly 10 billion US dollars in 2020.

Of course, more and more manufacturers also see this trend and enter the super-integration market, making the competition more fierce. The background of this inevitable trend comes from the qualitative improvement in the agility, ease of use and performance of the super-fusion architecture compared with the traditional three-tier IT architecture.

SmartX steadily overtook on the corner

SmartX, which has been focused on super-integration since 2013, also benefits from this disruptive market opportunity. So far, in addition to having the largest super-fusion case in China such as Unicom, we have also obtained financial and manufacturing head customers like Haier, Taikang Life Insurance and Guotai Junan, who used to use products from IT giants such as EMC, VMware and Dell, but now gradually begin to replace the original architecture with our super-fusion products, which are used not only in the development and test environment, but also in the production environment. To achieve this achievement comes not only from our focus in this direction, but also from the inevitability of super-integration to subvert the traditional IT architecture.

Another thing worth mentioning is that the 2018 Q1 IDC report shows that SmartX is already in the top three in the super-integrated market of the financial industry. The financial industry is the industry direction that we will further focus on. We firmly believe that in this market, we can most comprehensively test the reliability and value of products, and get the most representative user feedback. At the same time, some of the important product features of our 3.5 also come from the actual market demand in the financial sector.

About SMTX OS

First of all, SMTX OS is the trademark of our software products, which has been registered in China and major overseas countries. SMTX OS is currently the core software of the super-converged architecture and will be the core operating system of the enterprise cloud in the future. Although everyone is talking about the concept of a large and comprehensive enterprise cloud, we believe that just as Amazon's success started with "EC2", doing a good job of computing and storage is the first step to effectively build an enterprise cloud. At present, our main energy and resources are to build the "EC2" of this enterprise cloud.

Figure 1. SMTX OS will become a super converged enterprise cloud operating system in the future.

Why release SMTX OS 3.5

SMTX OS has gone through stages 1.0,2.0,3.0 before. 1.0 and 2.0 are our versions before 2017, and their highlight is distributed block storage under super convergence. The core of China's largest super-fusion case Unicom Voyun comes from version 2.0, and it should be said that the operation is very stable. The whole super-integration includes the expected elastic scalability, high performance, low latency and easy management of distributed storage. Have been fully reflected.

In version 3.0, a complete super-convergence product is built on the basis of the core distributed block storage, which enhances a lot of product content, including the enhancement of super-fusion at the virtualization level, including the support of more virtualization versions such as VMware and Xenserver. These super-fusion advanced ideas are all implemented in this version. Now many customers, such as financial customers and large manufacturing customers, are cut in through version 3.0.

Figure 2. Evolution of SMTX OS version

However, customers still have a question about whether the super-converged architecture can completely replace some of the existing traditional IT infrastructure architecture. Regardless of traditional architecture or super-converged architecture, users have the same needs when considering IT infrastructure, hoping to simplify the complexity of IT, reduce TCO, and enhance the agility of the entire IT infrastructure to cope with the rapid business growth in the fast-growing Internet era. In this way, we can shift more energy from the operation and maintenance and management of the original IT infrastructure to focus on the innovation and development of our own business. The current super-integration has been able to meet most of the above needs. However, for some core application scenarios, such as core databases and production systems, users want to achieve the RPO, RTO and disaster recovery capabilities that match the traditional IT solutions, but also maintain the advantages of super-integration, agility and ease of use.

What we want to do or the product goal is to help users make better use of the core business system in the super-converged architecture and get a better experience. On the one hand, you can enjoy the benefits of superconvergence itself, including simplifying IT and reducing TCO, while avoiding complex situations like the mixture of traditional architecture and superfusion architecture in the IT infrastructure.

With regard to disaster recovery capability, we can be divided into different levels from low to high, such as single disk failure recovery, single node disaster recovery, single failure, single data center and metro failure. For the core application system, of course, it is required to provide not only single-disk, single-node and single-rack disaster recovery capabilities, but also site-level and metro-level disaster recovery capabilities, which is also the focus of the functions to be released in 3.5. Next, I will introduce some of the more important features provided in 3.5 in turn.

Analysis of five characteristics of SMTX OS 3.5

Feature 1: double active cluster

First of all, the important point is the dual active cluster. We just mentioned that the IT infrastructure requires different disaster recovery protection. Double active cluster is actually a function introduced in SMTX OS 3.5 to deal with site-level failures and city-level site failures.

Let's take a look at the deployment model of a dual-active cluster. First of all, we need to deploy SMTX OS clusters in two different data centers in a city, and the RTT between the two sites is required to be less than 5 milliseconds. At the same time, we need to deploy the arbitration node in the third site, which is mainly used to resolve the arbitration when the IT system of the cluster has a brain crack and decide which site should be the only site that exists.

Achieving real-time synchronization on the underlying data is called double-live, which enables multiple data centers in the same city to achieve RPO=0, second-level RTO level guarantee. When one of the sites fails, the business or VM running above will be automatically restored to another site through the HA function provided by our system, so that business continuity can be achieved.

A very important point is emphasized here, which is also the theme that release 3.5 has been running through. The double-active or double-active function we provide does not only refer to the double-active function at the storage level. First of all, our dual-active cluster is completely dependent on the underlying dual-active storage function, that is, the data is between two different sites and can be synchronized in real time. More importantly, in addition to providing dual work at the storage level, it also provides site-level protection for the entire IT infrastructure. This IT infrastructure includes services supported by storage, virtualization, networking, monitoring deployment, and so on. In other words, we provide the dual-active function of the complete IT infrastructure, or site-level disaster recovery function, in a super-integrated software architecture.

When one of our sites fails, not only the data can guarantee RPO=0, second-level RTO level, VM can drift past. At the same time, our virtual machine management function, network management function, as well as disaster recovery, monitoring, alarm and other functions, can be automatically restored in another site. This is the huge benefit that we think super-integration brings. Traditional IT deployment, our virtualization, storage, network, including some supporting systems, each system may come from different manufacturers, each manufacturer's products need to have their own independent configuration or dual-live deployment requirements. If we want to achieve the dual activity of the entire IT infrastructure or site disaster recovery protection, we need to do independent configuration for each product. At the same time, because there are certain dependencies between products, it is very complicated when we want to use IT infrastructure as site protection. And because SMTX O S is a super-integrated architecture, which provides all the one-stop features, you only need to do a simple double-active configuration in our interface to achieve IT infrastructure disaster recovery protection, which realizes the simplification and enhancement of the core system experience on super-integration, which greatly reduces the usual operation and maintenance load of IT administrators.

Feature 2: data backup and remote disaster preparedness

The above is an introduction to our double active cluster, followed by data backup and remote disaster recovery. As mentioned just now, double active users actually refer to double active users in the same city. We have a strict requirement that the RTT between two sites must be less than 5 milliseconds. Because of this restriction, double active users can only do data protection between the same city. Sometimes enterprises have more remote disaster preparedness protection needs, at this time we provide in 3.5 data backup and remote disaster preparedness functions to meet the requirements.

First of all, let's take a look at the deployment model. At the time of deployment, we can deploy site An and site B in different cities, such as City An and City B. site A that we are now running in VMware is called a protection site, and site B is called a backup site. If we want to protect the business or virtualization of site A, we only need to select the objects that need to be protected by creating a protection plan on the administrative interface. The objects we need to protect can protect one or a group of protected objects. There are also two different types of protection objects, including user virtualization, namely VM. We can use VM as an example to protect the business, or we can choose independent virtual volumes as protection.

When we create a protection plan, we can select the backup time and backup interval after selecting the protection object. For example, we can provide a backup interval of at least one hour. For example, we can select site B as the target site for backup. In this way, we can complete the creation of the protection plan. When the protection plan is created, SMTX OS's system periodically faces VM at the protection site, or takes snapshots of selected protection objects. When the snapshot ends, the changed data is transferred incrementally and compressed to the backup site.

Suppose that when our site A fails, the user can choose to restore the VM at site B through the operation of the interface, so as to ensure business continuity. Because the interval between our backups is currently one hour, we are able to reach the hour level of RPO, while the overall RTO is at the minute level.

In addition to backing up the data, we can also support that when site A recovers, we can pull the data back from site B or synchronize it back to site A, because site B may be a disaster recovery site and resources are limited. When site An is restored, we naturally want to run the data on site A. We just need to synchronize the data back on the interface, and the business will be migrated to the protection site of site A.

Further, we support not only backup from site A to site B, but also backup from site B to site A, that is to say, our backup is bidirectional. In addition, there can be multi-site many-to-many backup, and a site A can be backed up to site B and site C at the same time to provide a higher level of disaster recovery protection.

In the data backup and remote disaster recovery functions we provide, we protect not only the data, but also take the business as the starting point to consider business continuity. When some configuration changes occur to our protected objects such as VM itself, the configuration changes are automatically synchronized to the backup site.

For example, suppose that when VM is used at site A, there are only two virtual volumes when the protection plan is initially created. As we continue to use and expand our business, we want to add one more virtual volume, that is, three virtual volumes, with some business running on it. At this time, in the process of the next data synchronization and backup, in addition to synchronizing the contents of the newly added virtual volume to the backup site, it will also synchronize the virtual operation change of the virtual machine to the remote site. when the remote site starts the virtual institution, three virtual disks are loaded at the same time, which avoids some data running on the new data disk, and there may be some problems when starting. This is also to maximize the continuity of the business, that is, to protect data from a business point of view.

This is the SMTX OS 3.5 cross-city data protection and disaster recovery function that we provide. These two functions from the current situation, only a few super-convergence manufacturers can have, and is completely different from the traditional storage vendor point of view of protection, we do data protection from the perspective of the entire IT infrastructure.

Feature 3: virtual machine management

Next, let's introduce virtualization, the three main modules of hyperconvergence: computing, storage, and networking. Computing refers to virtualization. If you have known us before, we still have a wealth of platform support for virtualization, including VMware, Xenserver platforms and embedded KVM platforms, which can provide good super-convergence deployment and support. Here more emphasis is placed on its own KVM-based embedded virtualization platform, which is highly customized and has done a lot of in-depth optimization and integration, providing what we think is the best ease of use and the best user experience.

The internal virtualization capabilities have also been greatly enhanced in version 3.5, which I will briefly introduce in four points.

The first is the ability to place groups. When the size of the cluster is relatively large, especially when users use core businesses, such as Oracle RAC scenarios, it actually requires special restrictions on the nodes on which each virtual machine or business runs. For example, the two VM of Oracle RAC want to run on different physical nodes, or some special requirements are running on the same physical nodes. In response to this requirement, we provide the ability to place groups to help you choose which node VM is running on according to the placement policy provided by the user.

This combines the automatic scheduling function we provide below, which means that when I create a VM, the user does not need to manually select or specify which node to run on, the system will automatically select according to two conditions. The first condition is the user-defined placement group function, depending on which placement group policy you belong to and how to define your placement rules. At the same time, combined with the current pressure of each node in the cluster, including the pressure of CPU, the pressure of memory and so on, I choose a node with as low load as possible to carry the current business.

The third function is the function of VM HA, as a virtualization management platform VM HA is a very basic function. In the previous version 3.0, we provided a guarantee of reliability that 5 minutes or so would automatically HA the virtualized business to another runnable node in the event of a failover. Further improvements have been made in version 3.5 to reduce the 5-minute limit to 3 minutes, which means that when your business fails, you will have a maximum of 3 minutes of failover time, and we have increased the overall speed by 40%.

Fourth, we also provide virtualization-based batch creation, as well as batch migration operations, which is also a further improvement in the user experience.

Feature 4: block storage service

Block storage service, before 3.5, block storage service only serves the virtualization services within the super-convergence, such as building a virtual machine on it, and the virtual machine can directly use the pooled storage capacity at the bottom of the super-convergence architecture. But in fact, we encounter a problem when communicating with customers. The environmental requirements of the application systems running in the customer's actual production environment are complex and diverse. For example, it is possible that some systems must run on physical machines, and some systems want to run in containers, such as K8S and Docker. Some are running in a virtualized environment, and some are running in a hyperfusion. Is it possible for us to uniformly manage and integrate the diverse needs of our customers, especially the storage requirements?

In order to meet the needs of customers, we provide block storage services in 3.5. as the name implies, the block storage capabilities provided by SMTX OS itself are exported through the iSCSI standard interface. In this way, for external physical machines, containers, K8S or other application environments, you only need to access the storage capacity in our super-converged cluster through the standard iSCSI protocol, which is called block storage service.

What are the benefits of block storage? First of all, we can see the most obvious, because the block storage function of SMTX OS itself is very powerful, including the data protection function. It has just been mentioned that it includes the protection function of single disk, single node and single rack, as well as the introduction of dual active function and cross-region disaster recovery protection function. These functions can continue to be reused, as long as the use of super-converged block storage services, you can also enjoy these storage functions. At the same time, SmartX distributed block storage is stable, simple, reliable, flexible and easy to expand, and these functions are also included in its own block storage service.

Figure 3 / 4. An OS integrates steady-state and sensitive services

Feature 5: new interface design and operation and maintenance function enhancement

Figure 5. SMTX OS 3.5 software interface

In addition, let's introduce the newly designed interface in version 3.5, as well as the more powerful and easy-to-use operation and maintenance functions. First of all, the very important function of multi-cluster management has been added. Many of our customers already have multiple clusters when using SMTX OS products, and most customers may have many clusters to manage at the same time. When he needs to manage multiple clusters, we provide multi-cluster management in 3.5, that is, all SMTX OS clusters can be managed at the same time in one interface.

At the same time, we provide a more advanced monitoring analysis and alarm function, in this version provides more than 200 computing, storage, network and related indicators monitoring and statistics, as well as changing trends. At the same time, TopK analysis is provided to quickly find out which virtual machines or data volumes are accessed frequently and which are hot virtualization. We make some planning or deployment according to statistics. We also provide customization of the monitoring panel, users can customize the monitoring panel and alarm properties they want to see, such as what conditions the alarm can be done according to.

There are also report exports that include hardware assets, which is a very practical feature, and many customers will encounter asset statistics problems. In addition, it can intelligently identify the node layout and the location of hardware disks, and support the visualization of third-party service topologies.

Modular authorization model adapts to wider hardware

The following is about License, if you are familiar with other products, like the general IT products when providing License is not so flexible, such as the basic version, enhanced version, enhanced version, etc., each different version contains a set of functions. When you want advanced features, you can't just pay for one feature, you have to pay for a set of features, there may be some features that you don't need. According to the needs of our customers, we provide a more flexible way of License authorization, that is, our overall License is licensed to the enterprise version, which contains all the basic features, including its own virtualization, network, storage and other functions. Advanced features include double activation, remote disaster recovery, and authorization through an independent License. For example, customers only need the protection of double active users, and you only need to pay for one function. If you only need the feature of remote disaster recovery, you only need to purchase the License of remote disaster recovery, and you do not need to be forced to sell you some features that you do not need. This is a very flexible License mode customized according to the needs of users.

Figure 6. Flexible license mode is widely adapted to the latest hardware platform

At the same time, we have adapted the hardware compatibility in this version, including mainstream hardware models, including some of the most mainstream models such as Lenovo, Tide, Dell, Chaowei, Huawei and Hewlett-Packard. From the perspective of hardware disks, Intel, Samsung and Toshiba are supported on SSD, and all the most mainstream SSD models can support it.

Summary of version 3.5

To sum up, we first enhanced the business functions from several angles, including virtualization, operation and Web interfaces. At the same time, we also introduced some new features, including block storage, dual active, and remote disaster recovery.

I would like to stress again here that the reason why we are able to launch these functions is, on the one hand, to provide a super-converged architecture, which can provide the best user experience and help users migrate their core production system to the super-converged architecture as much as possible. At the same time, because we have our own product research and development team, we are able to quickly push out these functions in such a short time, which is very different from other super-integration manufacturers.

Figure 7. SMTX OS 3.5 greatly expands the boundary of value

When we released 3.5, in fact, our product value boundary has been greatly expanded. We believe that the current SMTX OS 3.5 products are actually super-integrated products with global competitiveness. We support virtualization scenarios to bare metal scenarios, including steady-state, sensitive, single data center, multiple data centers, disaster recovery requirements, easy-to-use requirements, and so on.

Future Planning of SMTX OS

The above is our introduction to 3.5 products. In fact, we have some customers using 3.5 products so far. Our current R & D team has begun to develop and design further new features for future versions, including version 4.0 and later versions. The goal of development and design is also around the just mentioned so that users' core applications better in the super-integration in a better experience, this is our product always the same goal.

Figure 8. The future development of SMTX OS

You can take a look at the figure above. Our current product development idea is as follows: first, the bottom layer provides four core cornerstone modules, including block storage function above and virtualization function below. on the left is zero operation and maintenance management (very powerful automatic operation and maintenance management), as well as data protection and disaster recovery functions. With these four basic functions, the scenarios that can actually be covered include applications such as server virtualization, block storage, services such as VDI, ROBO, and core business scenarios. For the needs of users of some more advanced products, in fact, it will be developed to include the integration of backup ecology. We will open some storage access and backup interfaces. We support the integration of third-party manufacturers with SMTX OS products to back up the data in our super-fusion. At the same time, it will also provide some applications like secondary storage.

Secondly, there are some more core application scenarios, such as the core data user application scenario, which has a very high demand for performance, including that the trend of all-flash is becoming more and more obvious, so we will further guarantee SLA in all-flash and all-flash scenarios. And according to the development idea of enterprise cloud, it will provide advanced features such as multi-tenancy, self-service and so on, to help a large enterprise's IT infrastructure to do good operation and management.

Finally, we will further expand the entire ecology, including support for K8S containers, including providing Tencent App Center, and we also support third-party manufacturers to embed some of his applications into super-fusion architecture products. Through our cooperation with each other, we can bring users the most perfect super-integration experience.

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.