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

In this issue, the editor will bring you about how to analyze the SAP in the database definition storage. The article is rich in content and analyzed and described from a professional point of view. I hope you can get something after reading this article.

SAP

When it comes to in-memory databases, the first thing that comes to mind is SAP HANA. HANA (High-Performance Analytic Appliance) refers to high-performance analysis equipment. It is known at a glance that it is mainly used in real-time data analysis and is a combination of hardware and software. SAP officially released HANA software in 2010, but SAP has no hardware, so it must be provided together with hardware manufacturers. As the first commercial in-memory database product, SAP HANA has developed rapidly and become one of the fastest growth points of SAP. SAP HANA is not only suitable for OLAP business, but is also being promoted to OLTP scenarios.

You may ask, since SAP HANA is an in-memory database, what does it have to do with storage, because the data is calculated in memory?

The data of SAP main memory database is not only in memory, but also constantly written to the hard disk. SAP HANA usage storage is mainly reflected in the following aspects:

1. Save the installation file

2. Daily backup

3. Data preservation. The updated data in memory is saved to the hard disk every 5 minutes by default.

4. Redo log storage.

Therefore, the in-memory database also needs to be stored, but these requirements for storage is only a performance requirement, and the amount of data is generally small, so the server itself can basically meet the requirements by inserting some SSD disks. As a result, we see that many all-in-one SAP HANA machines, including Huawei servers, are also on the compatibility list.

However, in SAP HANA scale-out (scale out) scenarios, external storage is more appropriate. Let's first look at the principle of SAP HANA's Scale out.

SAP HANA supports multiple hosts to work at the same time to form a cluster, and there can be many hosts, such as 100,95 of which work and 5 do hot backup.

When any one of the working hosts fails, the backup host needs to replace the working host.

The SAP HANA architecture is logically a shared-nothing architecture, which means that each host has its own separate data and logs files. What to do when there is no data on the backup node? it can only be taken from the external storage. That is, all hosts can access a set of shared storage.

Host's own switching SPA HANA can be done on its own, but how to do data switching? How to ensure that the failed host does not write to the storage and release the stored resources to the new take-over host for exclusive use. In order to solve this coordination problem, SAP HANA developed a "Storage Connector API". Hardware vendors support this API, which can cooperate to realize the sharing and isolation of storage resources.

Let's look at the role of this API in two situations:

1. Sharing SAN

This kind of situation is widely used and the performance is good. However, because the host can only see its own data and logs files, if a host fails, the backup host needs to take over the data and logs files, and the failed host can no longer read or write, otherwise the file will be corrupted. At present, SAP HANA has realized that under the SUSE LUNIX environment, using the multi-path software of SLES (SUSE Linux Enterprise Server), the storage that supports SCSI-3 (using Persistent Reservations) can be switched and isolated. Since newer stores generally support the SCSI-3 command set, this mode supports the most vendors. Because there is no programming on the storage side, it can be certified by SAP HANA as long as it is tested. HDS VSP in high-end storage is currently on the certification list (with Hitachi blade servers). It is understood that if you do not use FC connections and use iSCSI, then the storage side may need some simple programming, such as writing scripts to support SAP HANA's "Storage Connector API".

2. Sharing NAS

If you use a shared file system, because interlocking is required, SAP recommends using a clustered file system, such as IBM GPFS. GPFS has its own mechanism for file sharing and isolation. If it is a normal NAS, such as the NFS file system, then SAP recommends that you develop the appropriate interface to work with Storage Connector API. When SAP HANA discovers the fault of a production host, it notifies the hardware management software, and then sends an instruction to reset the failed host, allowing the failed host to release storage resources.

You may want to ask, doesn't this also need the cooperation of the server? Because it's actually going to restart the server. This is true, and SAP HANA's scale out architecture authentication is not a separate authentication store, but is authenticated with the server. However, Watermelon Brother feels strange in this way, and it should not be the mainstream way of support.

In general, SAP HANA's main demand for storage is performance, while being able to support the switching of cluster hosts.

We found that because of the particularity of SAP HANA, there is no need for much coordination with storage, such as IUnip O priority, Cache coordination and so on, because when working, the data is basically in memory.

The above is how to analyze the SAP in the database definition storage shared by the editor. If you happen to have similar doubts, you might as well 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.