Oracle Analytics

Get Involved. Join the Conversation.

Topic

    Sathyanath Lappasi
    OAC Embedded BI Analytics
    Topic posted April 5, 2018 by Sathyanath Lappasi, last edited September 11, 2019 by Arijit ChakrabortyBronze Trophy: 5,000+ Points, tagged Reporting 
    74 Views, 4 Comments
    Title:
    OAC Embedded BI Analytics
    Summary:
    OAC Embedded BI Analytics
    Content:

    Implementation of OAC embedded analytics.

    Use Case: Embedded Analytics. (to embed all the visualizations of DVCS on to the external web portal of the customer through i-Frame approach.

    web portal Traffic: There will be around 10,000 concurrent unique visitors who will land on the page to perform actions such as analysis, look at the trends, slice and dice the data, apply filters and see the trends. They are basically viewers who will view and perform the above mentioned actions and they are are power users.

    Requirement and suggestions needed as per below questions:

    1. How many OAC instances need to be given for such use case?

    2. How many OAC (Enterprise) OCPUs need to be allocated?

    3. How many Database (Enterprise High Perf) OCPUs need to be allocated?

    4. How many Database instances need to be created?

    Comment

     

    • Gianni Ceresa

      The answer to 1-2-3-4 is ... depends ...

      You know that OAC is not just about concurrent users but also what you ask the system to do, what sources, where the load is etc.

      So the best matching answer is ... depends ...

      Good thing with OAC is that you can increase the power by switching to more resources if you see it doesn't follow anymore.

      PS: 10'000 concurrent users? Isn't that the maximum you expect with just few hundreds concurrent? So you expect to have around 10'000 guys at the same time doing analysis etc. ?

    • Sathyanath Lappasi

      Actual visitors landing on the page will be around 1,00,000. Customer expects a 10% concurrency. So 10,000 people will be visiting the page at the same time. This will increase over the time at 30% YoY.

      Data Source: My SQL(AWS) and few excel files

      Data Volume: 500 MB (excel + MySQL combined)

      For ex refer these links for embedded dashboards. consider such page gets 10,000 visitors across the world.:

      ***Moderator action (Timo): removed links to the external pages as they are considered as spam. If you want to show them as sample you can add an image.***

    • Gianni Ceresa

      I have the feeling you forgot a little detail ...

      OAC require a login to display anything, and I have the feeling you talk about a public website where you people to just freely see things. How did you planned to manage the authentication side of things?

      For the sizing it's quite impossible to give you an answer as in theory it will always work, slow maybe but still work. And the fact you use a MySQL hosted on AWS you can also have an issue on that side if you have 10'000 connections on the same time.

      Did you ask Oracle for some sizing guidance based on your use case (maybe telling them exactly the kind of models you have) ?

    • Sathyanath Lappasi

      yes I have and no response yet.

      Also Is there any limit for compute allocation. https://docs.oracle.com/en/cloud/get-started/subscriptions-cloud/mmocs/limits-service.html . if this is the case, how do we scale up on demand.

      I think data from any platform could be used no matter if it is on-prem or cloud (any platform) as it is gonna be a scheduled data sync and it gets loaded on to our platform and then on it will be seamless for any analytical task. Only challenge is the per user compute or memory allocation that I'm concerned of which may not be ignored when it comes to large audience base.

      I guess Files that stores these reports could be given guest access. still this needs to be validated.