BW on Hana , How About In The Cloud ?


I have been talking to customers on BW on Hana for a good while now – about their plans to make use of BW on Hana, and what roadblocks exist for its adoption etc. Here are the top three concerns that I hear the most

1. Projects to migrate to BW take a long time to complete, delaying value realized by business
2. Due to the length of the project and the cost to acquire HW, SW and Consulting services – the initial investment to be made looks high for some customers
3. Is it possible to try a few scenarios in BW on Hana quickly before going all in with a full migration?

Last year, several customers asked me why they should consider BW on Hana. These days, I hardly hear that question – I get more questions on “HOW” they should go about it . And to help customers and partners with this , my team and I have been traveling across the world doing hands on sessions on BW on Hana as well as BI on BWoH. Several other teams at SAP have also been doing similar enablement sessions, so I think the “HOW” part will be adequately taken care of too.

But we still need to better address the 2 questions on length of migration projects as well as the initial cost. And this why I wanted to ask your opinion of putting BW on Hana on cloud, on a subscription basis. There are many potential deployment options like I scribbled in these pictures on what systems can move to cloud.
BW on hana alone in cloud

BW and BI in cloud

everything in cloud

I I really have no idea whether SAP can actually do any of these ( it needs to make sense on technical and business dimensions). So please do not treat this as an official SAP statement. My intention is strictly to present my thought process to the ecosystem and get some feedback from all of you. And of course , this is just an “additional” deployment option, not a replacement of on premises BW on Hana which I fully expect to continue.

Back to the 2 common concerns – how do we solve that? I think a cloud option is probably one worth considering.  Here is what I think might work

1. SAP or a Partner acts as the cloud provider and signs a subscription contract with customer via SAP Hana Marketplace

2. Customer gives a copy of current BW system to the provider

3. Provider imports and migrates the copy into their cloud, and does testing. Maybe customer can do some user testing too.

4. With necessary signoffs, the solution is cut over into a productive instance on a cloud that customer can access

5. Any extra data/ transports etc that are needed are applied, deltas initialized, BI tools connected etc

6. Customer starts using the productive instance of BW on Hana with some agreed up on SLA

7. Customer starts paying the monthly subscription, with some provisions for data growth, additional BI clients and so on in future

I probably made it sound simpler than it is – but hopefully the general idea is clearly coming across to you.

I think there are many advantages in the cloud model

1. Customer deals with one provider alone – not one each for HW, SW, Consulting and Support.

2. Since one provider has full control over end to end experience on multiple customers – economies of scale should kick in and result in less time to do migration, and less cost than customer doing all the steps individually via multiple partners.

3. With subscription pricing, there is no big initial outlay for customers . Monthly payments are predictable for budgeting reasons etc .

Of course there are some trade offs too that need to be considered

1. There might be some reluctance to move a data warehouse to cloud for fear of security. However, this reluctance is definitely coming down – and VPN and other security measures can be put in place .

2. Network can become a constraint for loading and reporting. However, given the speed of BW on Hana on both reporting and loading are faster than classic BW, some of this can definitely be compensated. Also, not all loads are real time into BW . There is of course the option of moving the source systems also into the same cloud and reap even more benefits.

Now the million dollar question(s) – what do you folks think about this? Do you think this will help more customers derive value from BW on Hana ? What other factors should be considered from the customer’s perspective ?

And about the third concern of how to get a jump start on BW on Hana without going all in – Do you think customers will like to get a Hana One Premium type system for BWoH that they can run in parallel with existing BW systems for just a few scenarios ?

Lets discuss here in the comments section, and pls fill out the polls below


Advertisements

12 thoughts on “BW on Hana , How About In The Cloud ?

  1. Hi Vijay
    I’ve only one question
    All these scenarios are posible but if we want to create a Cloud Scenario, each customer needs a HANA Appliance for each Productive BW enviroment!
    So, when will be able different Productive enviroments in then same HANA Appliance??
    Or virtualize productive scenarios??

    Thanks!!

    Santi

  2. Hi Vijay,
    Although cost is very attractive factor in on-demand model, I think there are lot more concerns in its adoption.One of them being network connectivity issue…many a times while accessing cloud apps, I have encountered stability issues, where in I am unable to figure out if the issue is because of my internet speed or at cloud service provider.And sometimes it was that the server was brought down for maintenance and other times issue was with my network connectivity itself.Although strong online capabilities is a feature of any cloud app… I would also vouch for offline capabilities for atleast some of important tasks!
    On 3 different BW on HANA models depicted above, I would want to test the response time or performance of real-time in memory capabilities that HANA brings in each of the scenarios to make a decision or comment further.As per my experience, network delays would mask the real time capabilities of HANA and defeat the very purpose of HANA adoption when moved to cloud landscape.
    Cheers,
    Usha

  3. Vijay,
    I’m curious where this would fit into the of SAPs Hybrid on-premise / cloud concept ? is that assuming that the cloud portion is the HANA Cloud or Success factors ?
    Anyway, I immeadiately saw in your post an architechture that provides a managed and consistent pathway for a company migrating to Infrastructure in the Cloud (as opposed to using HANA Cloud or Success factors). My experience is with AWS, so bear with me if I propose anything thats proprietary to them but not other providers (i.e OpenStack, Rack mount etc). Very (very) simply, it is:
    * Implement a VPN link to the cloud provider (similar to the AWS VPC concept)
    * Build the HANA on BW on the cloud infrastructure,
    * Some use cases will load a LOT of data, so depending on the bandwidth required speed, use a Direct Connect approach – http://aws.amazon.com/directconnect/

    The only issue I see with the whole concept is that it makes the business more dependent on their infrastructure vendor. Conceptually, changing cloud vendors is simple – lift-and-drop – but as the volume of data and the relationships between components become more complex, doing within a non disruptive time frame becomes more and more difficult; Simple but not easy 🙂

    hth

    • Vijay,

      a Cloud scenario should definitely have a market and would increase adoption speed of the end users a couple of smaller details that I find would be very important:
      – Clear Migration path from Cloud to on-premise (should be very very simple and fast)
      – Cloud should be able to offer not only HANA DB but as well the BW NW Stack in order to have best performance (seen too many customers who put HANA and a “low-end” NW server…)
      – bandwidth between the two sites will be quite important. I am not so sure that a VPN over an Internet Connection really would deliver the quality results that would be expected from a high performance machine on the other side. If I am not mistaken in VPN connections you do not get an SLA …
      – Security: This is a point which will need some more thought. Just VPN does not seem to be strong enough for me. It will probably require strong encrytion on the Database itself. Further some countries, especially in Europe have very strict laws regarding where data can be stored (so called safe countries).

      Yet I would very much look forward to see this offering since it will clearly be a driver for adoption!

  4. Vijay,
    Great stuff. Everyone of your scenarios are in play today. IBM and many of the IaaS/PaaS providers can do each of these scenarios. Scenario 1 and 2 have the issue of needing to integrate via the WAN which can be problematic for either large size or high frequency of data transfers. Scenario 3 solves the problem by putting everything in the cloud.

    One of the biggest barriers today for implementing HANA on the cloud is that only non-PRD HANA runs on virtualized environments. I am aware HANA can handle multiple DB images in single box, but is not really cloud.

    I look at all of these as being all transition patterns as the SW world marches towards SaaS based solutions. I see the real threat to HANA the cloud native analytics from Google, Amazon, and others. Like HANA, many are in memory, but they are totally cloud native and API driven.

    There are companies coming into the fortune 500 that will have never bought a server, an OS, a DB, or SW. They’ve bought a business outcome via SaaS. It is unlikely that any of these will appeal to them. HANA multi-tenant cloud native would go a long way.

  5. Vijay – for customers that have gone the new install route, did any regret going this route that you are aware of? What were the tough lessons learnt? For us is was approach to transports. This is a LONG shot :), but are you perhaps aware of clients that follow the new install route, with an existing SEM-BCS production system in place? [we determined for BW 731 we need BCS 736, but are concerned about the transporting process especially here]…Johannes

    • I don’t know any one who regrets it, Johannes. It is a very useful strategy if the old BW system has a lot of objects and data that are obsolete. Not sure who amongst those have BCS though

  6. Hi Vijay, good one as always. We just completed a BW on HANA POC, and followed the new install route, as opposed to system copy that brings you PCA. Its horses for courses, but I really like the new install route, as you get to decide what you want to transport over from RDBMS BW to HANA BW [transports not as clean as usual albeit], and do not end up with all the junk that might be sitting in the RDBMS BW DEV workbench….Johannes [@lombardjohannes]

  7. Im sure there would be a market.

    A couple questions come to mind:

    1) in your “cloud” scenarios, will it be an elastic, scalable cloud or merely “hosting”

    2) initial load of a huge warehouse can be a challenge, and is usually best handled by shipping physical media to the cloud infrastructure provider, mounting and loading from a co-located server or storage device

    3) part of the beauty of HANA is the real time decision support it enables. However, that would imply that the cloud is kept up to date in near real time on transactions from source systems. Issues of network traffic and optimal transaction batch sizes might need to be worked out

    • All good questions, Rick
      1. Probably start with hosting as first step and then move on to an elastic model if there is enough demand. We need data to make the decision
      2. Yes…first load is via a physical copy on disk
      3. Agreed – we are studying this in detail

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s