Azure中断降低了StorSimple 8100 [英] Azure outage takes down StorSimple 8100
问题描述
我们有一台StorSimple 8100，其中本地使用了大约22 TB.上周，美国西部地区发生了一次Azure中断，这导致我们的StorSimple失去了与Azure的连接.发生这种情况时，我们的数据量下降，所有用户文件都变得不可访问. 一位支持工程师告诉我，如果丢失了Azure连接，它将按设计工作.我们确实将备份设置为每小时运行一次，并且在备份开始后就发生了卷断开连接.
在我看来，如果您的本地设备包含所有文件的副本，并且如果它与Azure的连接断开，则必须崩溃，因此这似乎是一个有问题的设计.我们购买StorSimple的原因之一是因为我们认为，如果有的话，我们将拥有冗余 是一次Azure故障.
我想知道是否有任何方法可以防止Azure连接断开时本地设备卷减少.如果不是，则发出警告会很有帮助，以便可以从容地降低卷而不是降低卷.
如果StorSimple生产设备上的云连接失败，则取决于状态您的设备上，可能会发生以下情况:
用于本地数据在您的设备上 : 在一段时间内，不会出现中断，并且将继续提供读取服务.但是，随着未完成IO的数量增加并超过限制，读取可能会开始失败.
取决于设备上的数据量，写入也会在云连接中断后的头几个小时内继续发生. 然后，如果云连接中断了几个小时，写操作将变慢，并最终开始失败. (设备上有临时存储空间，用于将数据推送到云中.发送数据时会清空此区域.如果连接 失败，该存储区域中的数据将不会被推送到云中，并且IO也会失败.)
StorSimple设备将本地层中的旧数据分层到Azure. 如果设备失去与Azure的连接，则数据将在 主机可以访问本地磁盘.但是，如果主机请求云中存在的数据块，则设备会重试获取数据. 如果无法在4分钟内检索到数据，则将音量/lun放入 离线状态. 如果在4分钟内建立了连接，则会从Azure提取数据并将其提供给主机.
如果正在进行云快照，则该工作一直尝试将数据发送到Azure，如果长时间中断，作业将在以下时间被取消 4个小时.
总可用容量包括可用于数据，元数据和缓冲区的容量.您可以在 8100 存储在本地固定卷中的数据未进行重复数据删除， 压缩的.但是，本地固定的卷的快照将被重复删除.由于本地固定的卷已完全配置，因此在配置过程中，某些来自分层卷的现有数据可能会推送到云中.您可以配置警报 使用此 链接.建议您参考 StorSimple本地固定卷的常见问题解答"部分.
-------------- -------------------------------------------------- ----------------------------------
如果此答案有帮助，点击"标记为答案"或 投票.要提供有关您的论坛体验的其他反馈，请单击 在这里
We have a StorSimple 8100 with about 22 TB in use locally. Last week there was an Azure outage in the West US region which resulted in our StorSimple losing connection to Azure. When that happened, our data volumes went down and all user files became inaccessible. I was told by a support engineer that it was working as designed if Azure connectivity is lost. We did have backups set to run hourly and the volume disconnection happened just after a backup started.
To me it seems like questionable design if you have a local appliance with a copy of all your files that has to crash if it loses connectivity to Azure. One of the reasons we bought the StorSimple was because we thought we would have redundancy if there was an Azure outage.
I'm wondering if there is any way to prevent the local appliance volumes from going down if Azure connectivity is lost. If not, a warning would be helpful so that the volumes could be taken down gracefully instead of going down hard.
If cloud connectivity fails on your StorSimple production device, then depending on the state of your device, the following can occur:
For the local data on your device: For some time, there will be no disruption and reads will continue to be served. However, as the number of outstanding IOs increases and exceeds a limit, the reads could start to fail.
Depending on the amount of data on your device, the writes will also continue to occur for the first few hours after the disruption in the cloud connectivity. The writes will then slow down and eventually start to fail if the cloud connectivity is disrupted for several hours. (There is temporary storage on the device for data that is to be pushed to the cloud. This area is flushed when the data is sent. If connectivity fails, data in this storage area will not be pushed to the cloud, and IO will fail.)
The StorSimple device tiers old data from the local tiers to Azure. If the device loses connection with Azure, the data on the local disks is accessible to hosts. However, if a block of data which is present on the cloud is requested by the host, the device retries getting the data. If the data cannot be retrieved within 4 minutes, the volume/lun will be put in an offline state. If the connection is established within the 4-minute period, the data is fetched from Azure and served to the host.
If a cloud snapshot is in progress the job keeps trying to send the data to Azure and if there is a prolonged outage, the job will be cancelled after 4 hours.
The total usable capacity includes the capacity available for data, metadata, and buffers. You can provision locally pinned volumes up to 8.5 TB on the8100 devices.Data stored in locally pinned volumes is not de-duplicated and compressed. However, snapshots of locally pinned volumes are de-duplicated. Because locally pinned volumes are fully provisioned, some existing data from tiered volumes might be pushed to the cloud during the provisioning process. You can configure alerts using this link . Suggest you to refer the FAQ section for StorSimple locally pinned volumes.
--------------------------------------------------------------------------------------------------
If this answer was helpful, click "Mark as Answer" or Up-Vote. To provide additional feedback on your forum experience, click here
这篇关于Azure中断降低了StorSimple 8100的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
