Jon Perryman wrote: >One very important detail I did not mention is the location of your data >in the cloud. You may connect to a cloud location thinking that is where >it will be stored. In order to be more efficient, some clouds may redirect >your request to a closer location. Potentially the country of the requester. >Why send the data halfway across the world when their cloud has a >location closer to the point of origin.
Mike Schwab wrote: > Most clouds store data in the nearest facility for reduced read write >time. Some clouds replicate to other sites. Have been outages when a >cloud site went down and the data was not available. Itschak Mugzach wrote: > If you use S3, you can specify which zone to use. I agree with Itschak, but it’s even better than that. ALL the major public commercial cloud providers offer cloud object storage services with selectable geographies. (Why was there a presumption they don’t offer geographic choices?) For example, here’s the menu for IBM Cloud’s Object Storage: https://cloud.ibm.com/objectstorage/create#pricing You can choose Cross Region, Regional, or Single Site. Each of these choices then offers various geographic choices. Cross Region keeps copies of your data in multiple data centers across a continent-sized area. Regional keeps copies of your data in multiple data centers across a country or large metro area. Single Site is just what it sounds like: one copy of your data (typically with versioning) in one site. You can provision more than one of these choices if you wish. Moreover, you’re not limited to one public commercial cloud, and you’re not limited to public commercial clouds. Cloud object storage APIs are reasonably well standardized, and you can have cloud object storage pools wherever you wish — across multiple public commercial clouds and/or private/on-premises cloud object storage pools, as you prefer. For example, including an ex-missile silo site if you want. David Jousma: >So the issue of using public cloud storage is a question you have >to answer for yourself. “How quickly do I need to be able to >restore?” If its TB of data, streaming in at network speed, that >could be days or weeks. Will you be out of business by then? It’s a possible consideration (weighed against various other considerations), but backups of z/VM *itself* aren’t typically that big. That’s one reason why I mentioned that you could view the IBM TS7700-based approach (with TS7700-to-TS7700 cross-site replication — what’s known as a Grid configuration — combined with a cloud object storage tier) as “cloud object storage caching.” So if there were a “small pipe” issue when recovering then that issue is partially or fully mitigated thanks to the TS7700’s own virtual tape storage in front of the cloud tier. Another possible approach is that you put your cloud object storage “on-premises” alongside the IBM Z machine, or even on the IBM Z machine. You can host a cloud object storage server on an IBM Z machine quite easily. Then your in-country TS7700’s cloud object storage tier is the remote cloud object storage server, alongside or on the out-of-country IBM Z machine. And then your recovery is via the remote TS7700 (alongside the DR machine) which is just pointing back to DR machine’s cloud object storage service, or the cloud object storage adjacent to the IBM Z machine. No “small pipe” problem with that! Jon Perryman wrote: >Googles cloud backup/recovery is very different from IBM z/OS.... You headed off on a tangent here that I don’t think I encouraged. I’m not sure what you’re referring to. >No IBM z system has cloud backup. You can't backup z/OS to >any other cloud than that provided by TS7700. Yes, you really can! There are software-only cloud object storage backup/restore solutions for z/OS. The two IBM products that are most directly relevant are: IBM Cloud Tape Connector for z/OS IBM Advanced Archive for DFSMShsm These products are available individually or in the IBM Advanced Storage Management Suite for z/OS license package. There’s a helpful YouTube video about these products here: https://www.youtube.com/watch?v=Inih7c4VeiQ Some other vendors also have products in this segment. As I mentioned, I’m not aware of any vendors that offer a pure software-based cloud object storage solution *for z/VM* backups/restores. IBM’s offering for z/VM (which also works with other operating systems) is the IBM TS7700 with its cloud object storage tier, in your choice of “baby” rack mount or factory frame form factors. But for z/OS (and Linux on IBM Z/LinuxONE) there are some pure software-based choices available too. Moreover, it’s possible to configure both z/OS and Linux on IBM Z/LinuxONE as cloud object storage *servers*. ————— Timothy Sipples Senior Architect Digital Assets, Industry Solutions, and Cybersecurity IBM zSystems/LinuxONE, Asia-Pacific [email protected] ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
