Shulou(Shulou.com)06/01 Report--

In this issue, the editor will bring you about how EMC redefines the next generation of high-end storage. The article is rich in content and analyzes and narrates it from a professional point of view. I hope you can get something after reading this article.

EMC WORLD 2015 is over, the evaluation of the new features of high-end storage VMAX3, and will XtremIO and DSSD replace VMAX3? Let's talk about these issues with you today.

In fact, there are still a lot of media reports about EMC WORLD 2015, and I have to say that the propaganda campaign of EMC is good. Therefore, in order to illustrate my point, I will only focus on it.

First of all, the theme of EMC WORLD 2015 is to redefine the next generation. Alas, this redefinition has been used by EMC, and HDS likes to use this word.

The focus is on high-end storage and how EMC redefines NEXT.

Let's first take a look at the sales of VMAX3 revealed by EMC. EMC said that more than 500 units have been shipped so far, with a total capacity of about 200PB, that is, more than 90% of each 400TB camera is equipped with flash memory and 7% of it uses eNAS.

Since the release of VMAX3, EMC has been stepping up the development of features. At present, except for mainframes, other features are relatively perfect. The following features seem to be supported except that double-active SRDF/Metro needs to wait until Q3.

The editor thinks that there are only two major updates to VMAX3 announced at this EMC WORLD conference, one is FAST.X and the other is SRDF/Metro.

FAST.X is actually an extension of the FTS (Federated Tiered Storage) feature of the previous generation of VMAX. The main function of FTS is similar to the heterogeneous virtualization advertised by HDS and Huawei, but it is said that FTS does not sell much in China, and it seems that EMC does not actively promote it. Anyway, VMAX has a high market share, so there is no need to take over other storage. O (∩ _ ∩) O ha!

However, the function of FAST.X is much more powerful than that of FTS heterogeneous virtualization, because it combines VMAX's SLO strategy and FAST and FTS, and the focus is no longer on vendor heterogeneity, but on product heterogeneity, and data flows freely according to policies.

It is different from the heterogeneous virtualization of other vendors in that it supports more layers and focuses on product heterogeneity rather than vendor heterogeneity.

Data ranges from the highest XtremIO all-flash arrays, VNX/ third-party arrays, Data Domian backup storage, and even cloud storage, depending on the level of service. That is, almost all kinds of storage forms are included.

No architecture is perfect. Each architecture has its own focus, and the greatness of FAST.X is to make data flow best suited to its lifecycle through the VMAX3 SLO strategy.

This vision looks beautiful, but it confuses the editor. What data does VMAX3 keep on its own? Is it possible that VMAX3 will not sell in the future? VMAX3 became a conductor, not a musician. It is suggested that EMC add its own level of VMAX3 to this diagram. I think it should be under XtremIO, which can be called performance / capacity balance layer, O (∩ _ ∩) O ha!

The second important update is double living. Ever since HDS first launched GAD at the high end, the editor suspected that VMAX3 might embed VPLEX to solve the dual active problem of the array. Unexpectedly, the editor's prediction was wrong, and EMC still intends to solve the double live problem based on its long-famous SRDF extension. It is normal to think about this idea now, because the reliability and quality of SRDF is higher than that of VPLEX, embedded VPLEX, although physically there is no need for independent gateways, but logically the same, all VPLEX O need to go through VPLEX, and the reliability and stability of VPLEX should not be as good as SRDF. Therefore, instead of taking the combinatorial approach of eNAS, EMC has actually developed a SRDF/Metro feature. This also makes it can not be quickly introduced to the market, can only be supported in Q3 this year. At that time, Huawei's high-end storage V3 will also begin to be supplied in large quantities, and Huawei's HyperMetro dual-active feature can also achieve dual-active without gateway. The double-active war of high-end storage is already on the brink of trigger.

EMC VMAX3 also has an update to support ViPR 2.2. That is, VMAX's SLO strategy can be managed through ViPR, which is good news for enterprises that plan to use ViPR for automated deployment.

Other updates have basically been mentioned before, so they will not be carried out.

Finally, it is just a personal guess to answer the question of how EMC will position high-end storage in the future.

From the current signs, the VMAX series is still the direction of high-end storage in the future. Why would you say that?

One theory is that the future is the world of all-flash arrays, and XtremIO will replace VMAX in the future. XtremIO is really gaining momentum, and according to the Gartner report, it is already the king of AFA in 2014.

And a lot of updates have been announced on EMC WORLD 2015.

However, when the editor saw that VMAX3's FAST.X included XtremIO in the performance layer, he understood that XtremIO would not replace VMAX. Because the bosom of VMAX3 is bigger than XtremIO, the pattern is also bigger than XtremIO.

Moreover, in the pursuit of reliability, VMAX3 is significantly higher than XtremIO several levels. For example, VMAX3 develops SRDF/Metro to achieve dual active users instead of embedding VPLEX. XtremIO still focuses on performance. For example, at the beginning of this year, the editor knew that XtremIO replication did not use SRDF code, but Recover Point. According to the newly released specifications, the asynchronous replication RPO of XtremIO can only reach the level of minutes (60s), while the SRDF of VMAX can achieve seconds (3-5s). The main reason why XtremIO does not achieve such a low RPO is to ensure the performance of AFA.

Therefore, the editor's observation is that VMAX3 places more emphasis on reliability, while XtremIO emphasizes performance. Therefore, XtremIO should not replace VMAX in the future.

What about DSSD? DSSD released more details at this EMC WORLD conference, thinking that DSSD might replace VMAX, but now it seems unlikely. Because DSSD takes a completely different path-- hardware definition.

Now the software definition is popular in the industry, but as EMC's VP said, the hardware definition is not out of date, just like the sine and cosine curves in the following figure, software definition and hardware definition will appear alternately, leading the change in storage.

Therefore, DSSD is a typical hardware-defined dedicated hardware product (of course, with the cooperation of underlying software), which is mainly aimed at the third platform applications, and is characterized by high-density flash storage close to memory latency. The key point is that it only cares about performance, so there are no data services, snapshots, replication, and so on. Therefore, it is impossible to replace VMAX in the future. Although VMAX3's FAST.X is powerful, it is not expected that DSSD will become one of the layers, because after virtualization, the advantage of latency is lost.

As the application of the third platform is still in its early stage in China, it is expected that DSSD will not have much impact on the domestic storage market for the time being. And DSSD is still under intense research and development, and will not be able to GA until the end of this year.

In summary, EMC's VMAX3 continues to represent EMC's evolution of the next generation of high-end storage. Judging from the current signs, EMC believes that the next generation of high-end storage is a reliable data service platform for enterprises. In Chinese, the next generation of high-end storage will be more of a handsome talent than just a general.

After seeing how EMC defines the next generation of high-end storage, let's mention the idol Moshe Yanai's InfiniBox, some of the changes to the third-generation high-end storage studied by him.

It is said that InfiniBox has sold 100 units with a total capacity of 100PB, with an average of each 1PB, which is higher than VMAX3. It can be seen that InfiniBox is better at the scene of big data.

InfiniBox now supports FC and iSCSI, supports NAS by the end of this year, and then will support object interfaces such as swift,S3. Mainframe support is also planned for this year.

Moreover, data re-deletion and compression are also under development. Infinidat said that the main purpose of re-deletion is not to reduce costs, but to improve performance, because their current costs are already very competitive.

As you can see, Moshe Yanai for the next generation of high-end storage is more like a versatile general of the big data era, to have performance, to be reliable, to be efficient and efficient, a machine can basically meet all the needs.

Generally speaking, the next generation of high-end storage development, the current focus is still on the EMC VMAX3 and HDS VSP G1000 trends, of course, the father of high-end storage InfiniBox should also pay attention.

Domestic high-end storage as a whole has just started, does not have the ability to lead the trend, but we see its strong learning ability, and there are still some local innovation. The development trend is good, and the released features are also very attractive, but we just don't know when users will be able to use them, unlike foreign manufacturers who will announce the delivery time of the features.

This is how the EMC shared by Xiaobian redefines the next generation of high-end storage. If you happen to have similar doubts, please refer to the above analysis to understand. If you want to know more about it, you are welcome to follow the industry information channel.

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.