Forum Discussion

rahulabcd's avatar
rahulabcd
Level 3
7 years ago

Cloud Catalyst support

Hi All,

We want to create a Cloud Catalyst Server and would like to know if we go for Linux RHEL Server for the same, can we build this in Azure Cloud? As in technotes and docuements, it is mentioned that it can either be NetBackup Appliance 5240 or Linux RHEL 7.3 or later with similar configuration, but no where is specified that RHEL Server can be Physical or Virtual and if Virtual can it be in Azure Cloud with similar configurations?

8.1.2
Linux
  • RiaanBadenhorst's avatar
    RiaanBadenhorst
    7 years ago

    Hi,

    As far as I understand the use case is usually to have the CC onsite, and it would store the deduplicated segments which is a simple put operation. What you're describing there would be a lot more back and forth communications as the onsite appliance and the CC in the cloud would need to communicate a lot more as it exchanges the data.

    The transfer between the two should be deduplicated but I just think that there would be a greater requirement or dependency on the bandwidht / latency etc if you configure it this way. Maybe someone from Veritas can comment.

    In addition to that, the HW requirement is quite large so it might have a big cost impact in the cloud?

  • RLeon's avatar
    RLeon
    Level 6
    7 years ago

    As Riaan puts it, the Cloud Catalyst is designed for the following architecture:
    OnPrem(Media Server --->CC--->) ------------------------------------>PublicCloud(----->Object Storage)

    ... and not this architecture you are suggesting:
    OnPrem(Media Server --->) ----------------------------------->PublicCloud(----->CC---->Object Storage)

    Even though both architectures would transmit data in deduplicated form to the cloud, the second one has to have a "workload" VM running in the cloud to terminate the data, before forwarding the data to the cloud storage.
    The CC is really for people who do not want to have a workload VM running in the cloud just to terminate deduplicated data because workload VMs are costly. They just want to put data in the cheap cloud storage directly.

    NetBackup Media Servers have been able to send data directly to public cloud object storage - such as AWS S3 and Azure Blob Stoarge - since many version ago with their cloud connectors, but they can only send data in non-deduplicated form, which is expensive.
    Some people would put a VM Media Server in the cloud to terminate the traffic, so that OnPrem Media Servers could send data in deduplicated form to the Cloud Media Server. But this had two problems:
    1. It is desirable to NOT put a workload VM in the cloud just to terminate data.
    2. When the workload VM receives the data, it has to either store the data inside itself, using its own VM storage (expensive), or forward the data to the cloud's object storage location (S3 or Blob), which is not a configuration Veritas officially supports (I.e., Veritas would not help you configure it because it is more of a cloud specific config).

    With the Cloud Catalyst way, multiple OnPrem Media Servers could send data to the OnPrem CC, which would then forward them directly to the cloud's object storage (S3 or Blob) in deduplicated form - all WITHOUT a VM running the cloud to terminate the traffic first.

    If for some reason you have to have a VM running in the cloud to terminate the traffic, then as explained above, you don't even have to use the CC, all you need is just a Media Server in the cloud - which btw is supported.

  • RiaanBadenhorst's avatar
    RiaanBadenhorst
    Level 6
    7 years ago
    Can you please explain why you want to create in the cloud as this sort of defeats the purpose but I’m interested to know what the overall architecture is.
    • rahulabcd's avatar
      rahulabcd
      Level 3
      7 years ago

      Hi Riaan,

      We want to build the Cloud Catalyst in Cloud itslef to reduce the DC footprint. Our ultimate goal is to send the deduplicated data to cloud using Cloud Catalyst. Can you please confirm, how this will defeat the purpose, if we build the Cloud Catalyst in Cloud itself, as the job is to send the deduplicated data to cloud storage. I understand the data would travel over available bandwidth to cloud catalyst, which will be hosted in cloud, but the cloud storage usage will be much less due to deduplication.

      I have gone through multiple technotes and blogs, but haven't come acroos any of the article suggesting Cloud Catalyst can or can't be build in cloud.

      • RiaanBadenhorst's avatar
        RiaanBadenhorst
        Level 6
        7 years ago

        Hi,

        As far as I understand the use case is usually to have the CC onsite, and it would store the deduplicated segments which is a simple put operation. What you're describing there would be a lot more back and forth communications as the onsite appliance and the CC in the cloud would need to communicate a lot more as it exchanges the data.

        The transfer between the two should be deduplicated but I just think that there would be a greater requirement or dependency on the bandwidht / latency etc if you configure it this way. Maybe someone from Veritas can comment.

        In addition to that, the HW requirement is quite large so it might have a big cost impact in the cloud?

  • rahulabcd's avatar
    rahulabcd
    Level 3
    7 years ago

    Thanks all, we have build the Cloud Catalyst as on-prem and will be using it to send dedup data to cloud storage.