Shared publicly  - 
 
SAP's announcement of hosted Business One at CeBIT was expected, but the pronouncement that B1 was now "multi-tenant" was not expected. That announcement did not make the press release, but it was mentioned in a number of articles quoting SAP at the time. That multi-tenancy proclamation set off a firestorm of blogs because multi-tenancy is a distinction that matters. Exactly how much it matters and what the definition means is another debate.

Bring on SAP's +Richard Duffy, who stepped into the fray with a blog comment on +Dennis Howlett's post (http://www.accmanpro.com/2012/03/13/why-the-sap-businessone-multi-tenancy-discussion-matters/). That discussion stopped when Richard stepped out on vacation, but upon return, Richard recorded a detailed 40 minute video about B1 and multi-tenancy that I combed through with a double strong mug of green tea by my side. I thought it might be useful to post a few of my comments here and see if Richard has any clarifications.

Before I dive into the weeds, I'll offer my verdict: SAP should have avoided bringing multi-tenancy into this discussion and simply talked in terms of reducing hosting costs for B1 partners that could be passed along to customers. The B1 "Cloud Control Console" sounds like a neat piece of functionality, but now the discussion has splintered into "does SAP understand the cloud/multi-tenancy?" debates. Analysts are distracted and customers don't care - at least not about the technical guts of this. Having said that, the technical distinctions are interesting so let's have a look-see.

Some key points from +Richard Duffy's video:

- in the B1 version of multi-tenancy, each "service unit" can host multiple tenants. However, each service unit is limited to a particular version of Business One. Also, each tenant in the service unit must share the same add-ons. Update: In Richard's response, he clarifies the add-on aspects: "Different tenants in the same service unit can have different add-ons running - as long as the core B1 version is the same and all the tenants using the add-on are on the same version of the add-on."

- Partners will likely be managing customers running on different versions of B1, so each version's tenants would be housed on a different "service unit". For software updates, those tenants on a same service unit could be upgraded at once, but each service unit would need to be upgraded separately.

- you must be running on B1 version 8.81 or later to be able to take advantage of B1's "multi-tenant" capabilities.

- each company (tenant) on a particular service unit has its own database. (Some purists would disqualify B1 from true multi-tenancy on that basis alone. Others would argue that the cost of maintaining individual databases goes against the most cost-effective virtues of multi-tenancy).

- each service unit resides on its own instance on Microsoft SQL Server, which can support multiple instances.

So is this multi-tenant? I suspect multi-tenant advocates will have different verdicts, but the real MT purists will have eyebrows at full raise. My view: because each B1 version must be supported by a separate service unit, this seems more like a series of multi-tenant clusters than true multi-tenancy. Some have said this resembles the "mega-tenant" approach SAP initially tried with Business ByDesign that was not viable. However in this case, with individual partners managing their own hosting environments, this technical arrangement may work out well enough.

There are a few questions I'd like to better understand from Richard:

- how many B1 customers are on 8.81 or higher compared to the total customer base? What is the upgrade path like for those not yet on the "cloud compliant" releases?

- what about the separate service units needed for different combinations of add-ons? Will this lead to a proliferation of separate service units or is there a way to limit the complexity?

- how many service units (instances) can realistically be supported on a SQL Server?

- what are the hosting costs improvements before and after these innovations? For 25 customers? 200? 500? Is there a point where the scale of savings gets more substantial, or is there a cost benefit from the beginning?

My view: with many "pure SaaS" providers, including ByDesign, moving upstream to bigger customers, there is a need for cloud solutions for smaller businesses. Whether B1 is truly cloud or multi-tenant is not the most important factor. Whether customers love the product and whether partners are able to offer it at competitive prices is.

SAP seems to have made significant effort to reduce the infrastructure costs for partners via some multi-tenant characteristics like shared application and database servers. I think it's better for SAP to avoid talking up the "B1 is multi-tenant" point, which is a debate they are not going to win with analysts. Why not simply say "we're reducing the costs of B1 hosting with the best techniques available, and the partners will be able to pass this savings and hosting flexibility onto our customers".

Over to you Richard, let me know if I missed something or need to correct anything.
4
2
Dennis Howlett's profile photoTim Guest's profile photoRichard Duffy's profile photoDennis Moore's profile photo
9 comments
 
Hi Jon,

Thanks for the thorough viewing of the video and taking the time to post the comments.

To your specific questions, heres my answers and sorry for the length but I'll quote the original question against each answer

Q: how many B1 customers are on 8.81 or higher compared to the total customer base? What is the upgrade path like for those not yet on the "cloud compliant" releases?
A: I'll come back to you with a % split if I can but bear in mind of our 35,000+ customers today, our expectation is not that there will be a "rush to the cloud" from them but if they do, the upgrade process is very simple and many of those customers on earlier releases are already doing that upgrade because of the incremental functional enhancements we delivered (other than architectural ones like "cloud" and HANA support for Analytics).
Heres a document that covers those enhancements http://slidesha.re/Hlc8F8
and a video demonstrating the Upgrade Process - http://youtu.be/G5cfeFfo19Y
Q:- what about the separate service units needed for different combinations of add-ons? Will this lead to a proliferation of separate service units or is there a way to limit the complexity?
A: Different tenants in the same service unit can have different add-ons running - as long as the core B1 version is the same and all the tenants using the add-on are on the same version of the add-on..
Thats the beauty of the architecture - only the Add-On code is stored in the SBO-Common and the record of which tenant is using which add on.
The data is all stored in the db thats linked to the tenants company data.

Q:- how many service units (instances) can realistically be supported on a SQL Server?
A:Not trying to use weasel wording but this really does depend on the no of users, volume of trx and complexity of trx that are being processed by each tenant as well as the way the SQL Server is configured - how much RAM is allocated, now many cores, disk infrastructure etc.
I will share some guidelines we have developed based on ,oad testing from the SAP Labs team in China as soon as I can - I am just doing a final review.

Q: what are the hosting costs improvements before and after these innovations? For 25 customers? 200? 500? Is there a point where the scale of savings gets more substantial, or is there a cost benefit from the beginning?
A: The cost savings are immediate because with the new architecture and "partitioning" of tenants ( tenant partitioning may be a better term than multi-tenanting to avoid the semantic discussions LOL ) as we can now support multiple tenants on a single SQL instance which we didnt do before.
Again the cost savings are really dependant on how the hoster chooses to design their infrastructure to handle the load...you (the hoster) can break it down I guess across a few key areas like $ return per CPU core, $return per Mb of RAM, $ return per Mb of storage, $ return per Mb of data transferred - these are just my suggested metrics as I am not in the hosting business but based on how the IAAS guys price up their infrastructure I imaging these are the things that they track a hard cost against in the data centre and of course $ return per SQL Server license or CAL/SAL.
Also load patternswill impact on the cost savings - if all my users are active at the same time I need more hardware allocated per service unit but if say I have clients on different time zones with different work loads then potentially I can get better hardware utilisation rates.
Again, it becomes a philosophical question, you design for peak load so this is where an experienced hoster works their magic to make their money. Our job isd to make it as easy as possible for them to do this.


So I hope this helps....perhaps as you said the "is it true multi-tenant" isnt the real question but whether or not customers will love the product and can it be offered at a competitive price - I think the answer to that is yes based on my experience of working with the solution and talking to partners and customers.

Thanks again for the feedback and I am looking forward to more discussions on the topic!

Richard
 
Sorry, one additional point I meant to make around the "rush to the cloud" - What we are now delivering is choice and for new customers they can now choose the solution based on the functionality and then choose on premise or on demand.
So I think the growth we are seeing with Business One will now accelerate because those customers that want an OnDemand solution can now take advantage of B1.
 
Thanks Jon.

I am also happy to do a Hang Out next week or some kind of open Q and A as well if you think it will help people get more clarity around the subject

Cheers


Richard
 
Does NetSuite, or Salesforce.com, or Workday run multiple versions of their software within the same exact infrastructure, database, etc.?
 
Hi Richard / Jon, Thanks for clarifying a few points I'm sure a few people had questions about. I think you are right Jon, all the end user will care about is does it work? How much does it cost? How quickly will they see an ROI? The rest is irrelevant to most people. Personally I think that if we see SAP Business One OnDemand running on HANA it will be very attractive and a good way to use the power of HANA without the capital outlay.
 
Simply putting B1 onto hosted/multi-tenant is SoSaaS as Phil Wainewright might say. SMEs are looking for something different.
 
Hi Dennis
What is the difference that SME's are looking for in your view?
The businesses I am talking to tell me they want flexibility, reliability, cost-effectiveness and a solution that their people like using.
Our existing customers tell us we are ticking those boxes with SAP Business One and our latest innovations like the cockpit, Analytics powered by HANA, mobility are delivering those.
I am intrigued by this concept of wanting something different - at the end of the day, all businesses need systems of record and systems of engagement that help them run their business and for small businesses (our main target market for B1) many of them just want a tool that works so they can get on with running the business that they do that makes them their money.
Our developers want to innovate for customers of all sizes and of course I am specifically interested in small businesses - so we are very interested in what the next solutions need to look like and I'd love to hear some feedback on this topic from everyone.
I have looked at a number of SaaS solutions and I am not seeing "revolution" - perhaps some folks out there in Google+ land are seeing more than me - if so I would love to hear what you think is not just "cool" and "different" but usable given the definition of what I am hearing small businesses want.
Or correct me if you feel I am wrong in saying that small businesses just want to run their businesses better so they can stick to what it is that they do best...and thats their core business.
Dennis, Jon - I am looking forward to our call next week with you and the team...we can toss around some answers to these questions perhaps?

Richard
 
Something that is transformational. I'm guessing B1 appeals to late adopters/laggards
Add a comment...