优化 Azure Linux VM 上的 MySQL 性能Optimize MySQL Performance on Azure Linux VMs

影响 Azure 上 MySQL 性能的因素有很多,主要体现在虚拟硬件选择和软件配置两个方面。There are many factors that affect MySQL performance on Azure, both in virtual hardware selection and software configuration. 本文重点介绍如何通过存储、系统和数据库配置优化性能。This article focuses on optimizing performance through storage, system, and database configurations.

Important

Azure 提供两个不同的部署模型用于创建和处理资源:Azure Resource Manager 模型和经典模型。Azure has two different deployment models for creating and working with resources: Azure Resource Manager and classic. 本文介绍使用经典部署模型的情况。This article covers using the classic deployment model. Azure 建议大多数新部署使用 Resource Manager 模型。Azure recommends that most new deployments use the Resource Manager model. 若要了解如何通过 Resource Manager 模型进行 Linux VM 优化,请参阅优化 Azure 上的 Linux VMFor information about Linux VM optimizations with the Resource Manager model, see Optimize your Linux VM on Azure.

自 2017 年 11 月 15 日起,仅在 Azure 门户中提供虚拟机。Starting November 15, 2017, virtual machines will be available only in the Azure portal.

利用 Azure 虚拟机上的 RAIDUtilize RAID on an Azure virtual machine

存储是影响云环境中的数据库性能的关键因素。Storage is the key factor that affects database performance in cloud environments. 与单个磁盘相比,RAID 可以通过并发访问提供更快的访问速度。Compared to a single disk, RAID can provide faster access via concurrency. 有关详细信息,请参阅 Standard RAID levels(标准 RAID 级别)。For more information, see Standard RAID levels.

通过 RAID 可改善 Azure 中的磁盘 I/O 吞吐量和 I/O 响应时间。Disk I/O throughput and I/O response time in Azure can be improved through RAID. 实验室测试表明,如果 RAID 磁盘数量翻倍(从 2 到 4,从 4 到 8,等等),磁盘 I/O 吞吐量将平均增加一倍,I/O 响应时间将平均缩短一半。Our lab tests show that disk I/O throughput can be doubled and I/O response time can be reduced by half on average when the number of RAID disks is doubled (from two to four, four to eight, etc.). 有关详细信息,请参阅 附录 ASee Appendix A for details.

增加 RAID 级别时,除了磁盘 I/O,MySQL 性能也会得到改善。In addition to disk I/O, MySQL performance improves when you increase the RAID level. 有关详细信息,请参阅 附录 BSee Appendix B for details.

可能还需要考虑区块大小。You might also want to consider the chunk size. 通常,区块越大,开销越低,对于大型写入操作尤其如此。In general, when you have a larger chunk size, you get lower overhead, especially for large writes. 不过,区块太大时,可能会有额外的开销,导致用户无法利用 RAID。However, when the chunk size is too large, it might add additional overhead that prevents you from taking advantage of RAID. 当前的默认大小为 512 KB,已被证明是大多数常见生产环境的最佳大小。The current default size is 512 KB, which is proven to be optimal for most general production environments. 有关详细信息,请参阅 附录 CSee Appendix C for details.

对于不同的虚拟机类型,可添加的磁盘数量存在多种限制。There are limits on how many disks you can add for different virtual machine types. Azure 的虚拟机和云服务大小中详细介绍了这些限制。These limits are detailed in Virtual machine and cloud service sizes for Azure. 可以选择设置磁盘较少的 RAID,不过,在本文的 RAID 示例中,需要附加 4 个数据磁盘。You will need four attached data disks to follow the RAID example in this article, although you can choose to set up RAID with fewer disks.

本文假定已经创建 Linux 虚拟机,并且安装和配置了 MYSQL。This article assumes you have already created a Linux virtual machine and have MYSQL installed and configured. 有关入门的详细信息,请参阅“如何在 Azure 上安装 MySQL”。For more information on getting started, see How to install MySQL on Azure.

在 Azure 上设置 RAIDSet up RAID on Azure

以下步骤介绍了如何使用 Azure 门户在 Azure 上创建 RAID。The following steps show how to create RAID on Azure by using the Azure portal. 也可以使用 Windows PowerShell 脚本设置 RAID。You can also set up RAID by using Windows PowerShell scripts. 在本示例中,我们将使用四个磁盘配置 RAID 0。In this example, we will configure RAID 0 with four disks.

向虚拟机添加数据磁盘Add a data disk to your virtual machine

在 Azure 门户中,转到仪表板并选择想要向其添加数据磁盘的虚拟机。In the Azure portal, go to the dashboard and select the virtual machine to which you want to add a data disk. 在本示例中,该虚拟机为 mysqlnode1。In this example, the virtual machine is mysqlnode1.

单击“磁盘”,然后单击“附加新磁盘”。Click Disks and then click Attach New.

虚拟机添加磁盘

创建一个新的 500 GB 磁盘。Create a new 500 GB disk. 请确保“主机缓存首选项”设置为“无”。Make sure that Host Cache Preference is set to None. 完成后,单击“确定”。When you're finished, click OK.

附加空磁盘

这会向虚拟机添加一个空磁盘。This adds one empty disk into your virtual machine. 再重复此步骤三次,以便为 RAID 附加 4 个数据磁盘。Repeat this step three more times so that you have four data disks for RAID.

通过查看内核消息日志,可以看到虚拟机中添加的驱动器。You can see the added drives in the virtual machine by looking at the kernel message log. 例如,若要在 Ubuntu 上查看,请使用以下命令:For example, to see this on Ubuntu, use the following command:

sudo grep SCSI /var/log/dmesg

创建具有更多磁盘的 RAIDCreate RAID with the additional disks

以下步骤介绍如何在 Linux 上配置软件 RAIDThe following steps describe how to configure software RAID on Linux.

Note

如果使用的是 XFS 文件系统,请在创建 RAID 后执行以下步骤。If you are using the XFS file system, execute the following steps after you have created RAID.

若要在 Debian、Ubuntu 或 Linux Mint 上安装 XFS,请使用以下命令:To install XFS on Debian, Ubuntu, or Linux Mint, use the following command:

apt-get -y install xfsprogs  

若要在 Fedora、CentOS 或 RHEL 上安装 XFS,请使用以下命令:To install XFS on Fedora, CentOS, or RHEL, use the following command:

yum -y install xfsprogs  xfsdump

设置新的存储路径Set up a new storage path

使用以下命令设置新的存储路径:Use the following command to set up a new storage path:

root@mysqlnode1:~# mkdir -p /RAID0/mysql

将原始数据复制到新的存储路径Copy the original data to the new storage path

使用以下命令将数据复制到新的存储路径:Use the following command to copy data to the new storage path:

root@mysqlnode1:~# cp -rp /var/lib/mysql/* /RAID0/mysql/

修改权限以便 MySQL 访问(读取和写入)数据磁盘Modify permissions so MySQL can access (read and write) the data disk

使用以下命令修改权限:Use the following command to modify permissions:

root@mysqlnode1:~# chown -R mysql.mysql /RAID0/mysql && chmod -R 755 /RAID0/mysql

调整磁盘 I/O 计划算法Adjust the disk I/O scheduling algorithm

Linux 实现了四种类型的 I/O 计划算法:Linux implements four types of I/O scheduling algorithms:

  • NOOP 算法(无操作)NOOP algorithm (No Operation)
  • 截止时间算法(截止时间)Deadline algorithm (Deadline)
  • 完全公平队列算法 (CFQ)Completely fair queuing algorithm (CFQ)
  • 预算期算法(预测)Budget period algorithm (Anticipatory)

在不同的情况下,可以选择使用不同的 I/O 计划程序来优化性能。You can select different I/O schedulers under different scenarios to optimize performance. 在完全随机的访问环境中,CFQ 算法和截止时间算法对性能的影响区别不大。In a completely random access environment, there is not a significant difference between the CFQ and Deadline algorithms for performance. 为保持稳定性,建议将 MySQL 数据库环境设置为截止时间算法。We recommend that you set the MySQL database environment to Deadline for stability. 如果有大量的顺序 I/O,CFQ 可能会降低磁盘 I/O 性能。If there is a lot of sequential I/O, CFQ might reduce disk I/O performance.

对于 SSD 和其他设备,NOOP 或截止时间算法可以比默认计划程序实现更好的性能。For SSD and other equipment, NOOP or Deadline can achieve better performance than the default scheduler.

在内核 2.5 之前,默认 I/O 计划算法为“截止时间”。Prior to the kernel 2.5, the default I/O scheduling algorithm is Deadline. 从内核 2.6.18 开始,CFQ 成为默认 I/O 计划算法。Starting with the kernel 2.6.18, CFQ became the default I/O scheduling algorithm. 可以在内核启动时指定此设置,或者在系统运行时动态修改此设置。You can specify this setting at kernel boot time or dynamically modify this setting when the system is running.

下面的示例演示如何在 Debian 分发系列中检查默认计划程序并将其设置为 NOOP 算法。The following example demonstrates how to check and set the default scheduler to the NOOP algorithm in the Debian distribution family.

查看当前的 I/O 调度器View the current I/O scheduler

若要查看计划程序,请运行以下命令:To view the scheduler run the following command:

root@mysqlnode1:~# cat /sys/block/sda/queue/scheduler

此时会显示以下输出,指示当前的计划程序:You will see following output, which indicates the current scheduler:

noop [deadline] cfq

更改 I/O 调度算法的当前设备 (/dev/sda)Change the current device (/dev/sda) of the I/O scheduling algorithm

运行以下命令以更改当前设备:Run the following commands to change the current device:

azureuser@mysqlnode1:~$ sudo su -
root@mysqlnode1:~# echo "noop" >/sys/block/sda/queue/scheduler
root@mysqlnode1:~# sed -i 's/GRUB_CMDLINE_LINUX=""/GRUB_CMDLINE_LINUX_DEFAULT="quiet splash elevator=noop"/g' /etc/default/grub
root@mysqlnode1:~# update-grub

Note

对 /dev/sda 单独进行此设置毫无用处。Setting this for /dev/sda alone is not useful. 必须对数据库所在的所有数据磁盘进行设置。It must be set on all data disks where the database resides.

此时会显示以下输出,指示已成功重新生成 grub.cfg 并且默认计划程序已更新为 NOOP:You should see the following output, indicating that grub.cfg has been rebuilt successfully and that the default scheduler has been updated to NOOP:

Generating grub configuration file ...
Found linux image: /boot/vmlinuz-3.13.0-34-generic
Found initrd image: /boot/initrd.img-3.13.0-34-generic
Found linux image: /boot/vmlinuz-3.13.0-32-generic
Found initrd image: /boot/initrd.img-3.13.0-32-generic
Found memtest86+ image: /memtest86+.elf
Found memtest86+ image: /memtest86+.bin
done

配置系统文件操作设置Configure system file operations settings

最佳做法之一是禁用文件系统上的 atime 日志记录功能。One best practice is to disable the atime logging feature on the file system. Atime 指文件的最后一次访问时间。Atime is the last file access time. 无论何时访问文件,文件系统都会在日志中记录时间戳。Whenever a file is accessed, the file system records the timestamp in the log. 但是,很少使用此信息。However, this information is rarely used. 如果不需要,可以禁用它,这样可以减少总体磁盘访问时间。You can disable it if you don't need it, which will reduce overall disk access time.

若要禁用 atime 日志记录,需修改文件系统配置文件 /etc/ fstab 并添加“noatime” 选项。To disable atime logging, you need to modify the file system configuration file /etc/ fstab and add the noatime option.

例如,编辑 vim /etc/fstab 文件,按以下示例所示添加 noatime:For example, edit the vim /etc/fstab file, adding the noatime as shown in the following sample:

# CLOUD_IMG: This file was created/modified by the Cloud Image build process
UUID=3cc98c06-d649-432d-81df-6dcd2a584d41       /        ext4   defaults,discard        0 0
#Add the "noatime" option below to disable atime logging
UUID="431b1e78-8226-43ec-9460-514a9adf060e"     /RAID0   xfs   defaults,nobootwait, noatime 0 0
/dev/sdb1       /mnt    auto    defaults,nobootwait,comment=cloudconfig 0       2

然后,使用以下命令重新装载文件系统:Then, remount the file system with the following command:

mount -o remount /RAID0

测试修改后的结果。Test the modified result. 修改测试文件时,系统不会更新访问时间。When you modify the test file, the access time is not updated. 以下示例展现了代码在修改前后的样子。The following examples show what the code looks like before and after modification.

之前:Before:

进行访问权限修改之前的代码

之后:After:

进行访问权限修改之后的代码

增加系统句柄的最大数量以实现高并发Increase the maximum number of system handles for high concurrency

MySQL 是高并发数据库。MySQL is a high concurrency database. 对于 Linux,默认的并发句柄数量是 1024 个,这个数量有时候不够用。The default number of concurrent handles is 1024 for Linux, which is not always sufficient. 通过执行以下步骤来增加系统的最大并发句柄数,以此支持 MySQL 的高并发。Use the following steps to increase the maximum concurrent handles of the system to support high concurrency of MySQL.

修改 limits.conf 文件Modify the limits.conf file

若要增加允许的最大并发句柄数,可在 /etc/security/limits.conf 文件中添加以下四行。To increase the maximum allowed concurrent handles, add the following four lines in the /etc/security/limits.conf file. 请注意,65536 是系统可以支持的最大数量。Note that 65536 is the maximum number that the system can support.

* <span data-ttu-id="a6f8e-203">soft nofile 65536</span><span class="sxs-lookup"><span data-stu-id="a6f8e-203">soft nofile 65536</span></span>
* <span data-ttu-id="a6f8e-204">hard nofile 65536</span><span class="sxs-lookup"><span data-stu-id="a6f8e-204">hard nofile 65536</span></span>
* <span data-ttu-id="a6f8e-205">soft nproc 65536</span><span class="sxs-lookup"><span data-stu-id="a6f8e-205">soft nproc 65536</span></span>
* <span data-ttu-id="a6f8e-206">hard nproc 65536</span><span class="sxs-lookup"><span data-stu-id="a6f8e-206">hard nproc 65536</span></span>

更新系统以实施新限制Update the system for the new limits

若要更新系统,请运行以下命令:To update the system, run the following commands:

ulimit -SHn 65536
ulimit -SHu 65536

确保在启动时更新限制Ensure that the limits are updated at boot time

在 /etc/rc.local 文件中放置以下启动命令,以便其在启动时生效。Put the following startup commands in the /etc/rc.local file so it will take effect at boot time.

echo "ulimit -SHn 65536" >>/etc/rc.local
echo "ulimit -SHu 65536" >>/etc/rc.local

MySQL 数据库优化MySQL database optimization

若要在 Azure 上配置 MySQL,可以使用本地计算机上使用的同一性能优化策略。To configure MySQL on Azure, you can use the same performance-tuning strategy you use on an on-premises machine.

主要 I/O 优化规则包括:The main I/O optimization rules are:

  • 增加缓存大小。Increase the cache size.
  • 减少 I/O 响应时间。Reduce I/O response time.

若要优化 MySQL 服务器设置,可以更新 my.cnf 文件,它是服务器和客户端计算机的默认配置文件。To optimize MySQL server settings, you can update the my.cnf file, which is the default configuration file for both server and client computers.

以下配置项是影响 MySQL 性能的主要因素:The following configuration items are the main factors that affect MySQL performance:

  • innodb_buffer_pool_size:缓冲池包含缓冲数据和索引。innodb_buffer_pool_size: The buffer pool contains buffered data and the index. 此值通常设置为物理内存的 70%。This is usually set to 70 percent of physical memory.
  • innodb_log_file_size:这是重做日志大小。innodb_log_file_size: This is the redo log size. 重做日志用于确保写入操作快速、可靠并且可在出现故障后恢复。You use redo logs to ensure that write operations are fast, reliable, and recoverable after a crash. 此值设置为 512 MB,这将为记录写入操作提供大量空间。This is set to 512 MB, which will give you plenty of space for logging write operations.
  • max_connections:应用程序有时候无法正常关闭连接。max_connections: Sometimes applications do not close connections properly. 值越大,服务器就有越多时间回收空闲的连接。A larger value will give the server more time to recycle idled connections. 最大连接数为 10,000,但建议的最大值为 5,000。The maximum number of connections is 10,000, but the recommended maximum is 5,000.
  • Innodb_file_per_table:此设置可允许或禁止 InnoDB 将表存储在单独的文件中。Innodb_file_per_table: This setting enables or disables the ability of InnoDB to store tables in separate files. 启用该选项可确保有效地应用多项高级管理操作。Turn on the option to ensure that several advanced administration operations can be applied efficiently. 从性能角度来看,它可以提高表空间传输的速度和优化碎片管理性能。From a performance point of view, it can speed up the table space transmission and optimize the debris management performance. 此选项的推荐设置是“开启”。The recommended setting for this option is ON.
    从 MySQL 5.6 开始,默认设置为“开启”,因此不需要任何操作。From MySQL 5.6, the default setting is ON, so no action is required. 对于早期版本,默认设置为“关闭”。For earlier versions, the default setting is OFF. 应在加载数据之前更改此设置,因为只有新创建的表才会受影响。The setting should be changed before data is loaded, because only newly created tables are affected.
  • innodb_flush_log_at_trx_commit:默认值为 1,范围设置为 0~2。innodb_flush_log_at_trx_commit: The default value is 1, with the scope set to 0~2. 默认值是最适合独立 MySQL DB 的选项。The default value is the most suitable option for standalone MySQL DB. 设置 2 支持最大程度的数据完整性,适用于 MySQL 群集中的主节点。The setting of 2 enables the most data integrity and is suitable for Master in MySQL Cluster. 设置 0 允许数据丢失,这可能会影响可靠性(但在某些情况下能提供更好的性能),适用于 MySQL 群集中的从属节点。The setting of 0 allows data loss, which can affect reliability (in some cases with better performance), and is suitable for Slave in MySQL Cluster.
  • Innodb_log_buffer_size:借助日志缓冲区,即使在事务提交之前未将日志刷新到磁盘,事务也可以运行。Innodb_log_buffer_size: The log buffer allows transactions to run without having to flush the log to disk before the transactions commit. 但是,如果有大型的二进制对象或文本字段,将很快地耗尽缓存,并触发频繁的磁盘 I/O。However, if there is large binary object or text field, the cache will be consumed quickly and frequent disk I/O will be triggered. 如果 Innodb_log_waits 状态变量不为 0,最好增加缓冲区大小。It is better increase the buffer size if Innodb_log_waits state variable is not 0.
  • query_cache_size:最好是从一开始就禁用。query_cache_size: The best option is to disable it from the outset. 将 query_cache_size 设置为 0(这是 MySQL 5.6 中的默认设置)并使用其他方法来提高查询速度。Set query_cache_size to 0 (this is the default setting in MySQL 5.6) and use other methods to speed up queries.

请参阅 附录 D ,了解优化前后的性能比较情况。See Appendix D for a comparison of performance before and after the optimization.

启用 MySQL 慢查询日志以进行性能瓶颈分析Turn on the MySQL slow query log for analyzing the performance bottleneck

MySQL 慢查询日志有助于识别 MySQL 的慢查询。The MySQL slow query log can help you identify the slow queries for MySQL. 在启用 MySQL 慢查询日志后,可以使用 mysqldumpslow 等 MySQL 工具来识别性能瓶颈。After enabling the MySQL slow query log, you can use MySQL tools like mysqldumpslow to identify the performance bottleneck.

默认情况下,此项未启用。By default, this is not enabled. 启用慢查询日志可能会占用一些 CPU 资源。Turning on the slow query log might consume some CPU resources. 建议临时启用此选项来排查性能瓶颈问题。We recommend that you enable this temporarily for troubleshooting performance bottlenecks. 若要启用慢查询日志,请执行以下操作:To turn on the slow query log:

  1. 通过在末尾添加以下行来修改 my.cnf 文件:Modify the my.cnf file by adding the following lines to the end:

     long_query_time = 2
     slow_query_log = 1
     slow_query_log_file = /RAID0/mysql/mysql-slow.log
    
  2. 重新启动 MySQL Server。Restart the MySQL server.

     service  mysql  restart
    
  3. 使用 show 命令检查该设置是否生效。Check whether the setting is taking effect by using the show command.

Slow-query-log ON

Slow-query-log 结果

在本示例中,可以看到慢查询功能已启用。In this example, you can see that the slow query feature has been turned on. 然后可以使用 mysqldumpslow 工具来确定性能瓶颈和优化性能,比如添加索引。You can then use the mysqldumpslow tool to determine performance bottlenecks and optimize performance, such as adding indexes.

附录Appendices

以下为在目标实验室环境中生成的性能测试数据示例。The following are sample performance test data produced in a targeted lab environment. 这些数据提供了在使用不同性能调优方法的情况下,性能数据趋势的常规背景。They provide general background on the performance data trend with different performance tuning approaches. 结果可能因环境或产品版本而异。The results might vary under different environment or product versions.

附录 AAppendix A

不同 RAID 级别的磁盘性能 (IOPS)Disk performance (IOPS) with different RAID levels

不同 RAID 级别的磁盘 IOPS

测试命令Test commands

fio -filename=/path/test -iodepth=64 -ioengine=libaio -direct=1 -rw=randwrite -bs=4k -size=5G -numjobs=64 -runtime=30 -group_reporting -name=test-randwrite

Note

此测试的工作负荷使用 64 个线程,尝试达到 RAID 的上限。The workload of this test uses 64 threads, trying to reach the upper limit of RAID.

附录 BAppendix B

不同 RAID 级别的 MySQL 性能(吞吐量)比较 MySQL performance (throughput) comparison with different RAID levels
(XFS 文件系统)(XFS file system)

不同 RAID 级别的 MySQL 性能比较
不同 RAID 级别的 MySQL 性能比较

测试命令Test commands

mysqlslap -p0ps.123 --concurrency=2 --iterations=1 --number-int-cols=10 --number-char-cols=10 -a --auto-generate-sql-guid-primary --number-of-queries=10000 --auto-generate-sql-load-type=write -engine=innodb

不同 RAID 级别的 MySQL 性能 (OLTP) 比较MySQL performance (OLTP) comparison with different RAID levels
不同 RAID 级别的 MySQL 性能 (OLTP) 比较MySQL performance (OLTP) comparison with different RAID levels

测试命令Test commands

time sysbench --test=oltp --db-driver=mysql --mysql-user=root --mysql-password=0ps.123  --mysql-table-engine=innodb --mysql-host=127.0.0.1 --mysql-port=3306 --mysql-socket=/var/run/mysqld/mysqld.sock --mysql-db=test --oltp-table-size=1000000 prepare

附录 CAppendix C

不同区块大小的磁盘性能 (IOPS) 比较Disk performance (IOPS) comparison for different chunk sizes
(XFS 文件系统)(XFS file system)

测试命令Test commands

fio -filename=/path/test -iodepth=64 -ioengine=libaio -direct=1 -rw=randwrite -bs=4k -size=30G -numjobs=64 -runtime=30 -group_reporting -name=test-randwrite
fio -filename=/path/test -iodepth=64 -ioengine=libaio -direct=1 -rw=randwrite -bs=4k -size=1G -numjobs=64 -runtime=30 -group_reporting -name=test-randwrite  

用于此测试的文件大小分别为 30 GB 和 1 GB,并且使用的是 RAID 0(4 个磁盘)XFS 文件系统。The file sizes used for this testing are 30 GB and 1 GB, respectively, with RAID 0 (4 disks) XFS file system.

附录 DAppendix D

优化前和优化后的 MySQL 性能(吞吐量)比较MySQL performance (throughput) comparison before and after optimization
(XFS 文件系统)(XFS File System)

优化前和优化后的 MySQL 性能(吞吐量)比较

测试命令Test commands

mysqlslap -p0ps.123 --concurrency=2 --iterations=1 --number-int-cols=10 --number-char-cols=10 -a --auto-generate-sql-guid-primary --number-of-queries=10000 --auto-generate-sql-load-type=write -engine=innodb,misam

默认值和优化值的配置设置如下所示:The configuration setting for default and optimization is as follows:

ParametersParameters 默认Default 优化Optimization
innodb_buffer_pool_sizeinnodb_buffer_pool_size None 7 GB7 GB
innodb_log_file_sizeinnodb_log_file_size 5 MB5 MB 512 MB512 MB
max_connectionsmax_connections 100100 50005000
innodb_file_per_tableinnodb_file_per_table 00 11
innodb_flush_log_at_trx_commitinnodb_flush_log_at_trx_commit 11 22
innodb_log_buffer_sizeinnodb_log_buffer_size 8 MB8 MB 128 MB128 MB
query_cache_sizequery_cache_size 16 MB16 MB 00

有关更详细的优化配置参数,请参阅 MySQL 正式说明For more detailed optimization configuration parameters, refer to the MySQL official instructions.

测试环境Test environment

硬件Hardware 详细信息Details
CPUCPU AMD Opteron(tm) 处理器 4171 HE/4 核AMD Opteron(tm) Processor 4171 HE/4 cores
内存Memory 14 GB14 GB
磁盘Disk 10 GB/磁盘10 GB/disk
操作系统OS Ubuntu 14.04.1 LTSUbuntu 14.04.1 LTS