Introduction

This document aims at helping position the different implementation options available for Xcelsius dashboards presenting BW data. There is currently 4 options : QaaWS, Live Office, BICS and LCDS. No option fits all needs and each option may be relevant depending on the customer's requirements. This article compares each option against others, and highlights pros and cons from a functionality, deployment, performance and implementation cost perspectives.

Deployment options

Functionality comparison table

Legend
functionality is natively available
functionality not available
functionality partially available or difficult to implement

Details

Security

All options allow end to end SSO using SAP credentials.

• With BICS, that is in fact the only option as it is using the Xcelsius dashboard is wrapped by BEx Web Runtime.
• With QaaWS and LO, it requires extra configuration within BOE but works well. Other options are available but are unlikely to be used with BW customers.
• The LCDS option is also relying on BOE so it is possible, but the whole authentication piece has to be coded using the BOE SDK.

Back to comparison table

Sharing and broadcast

• All options rely on a web service and therefore can be integrated into any portal with URL hyperlinking.
• The next obvious question is SSO between the portal and the backend system :
  • QaaWS and LO rely on BOE and therefore have native SSO to BOE Infoview
  • BICS use BEx Web runtime and has native SSO to NetWeaver portal
  • Any other option or combination will require additional configuration to get SSO between the portal and the backend.

Back to comparison table

Data sources

One of the main advantages of the Live Office option is the ability to leverage scheduled instances of Web Intelligence documents or Crystal Reports and keep the row level security. This allows an obvious performance gain and takes adhoc load off BW. This option is very relevant if the data loads are done once a day in BW. To achieve this, you will use multiple pass report bursting in BOE to schedule on behalf of users ("Schedule For" option in the CMC); and have the Live Office connection defined to refresh off the latest instance owned by the current user. You can also leverage scheduled reports with the LCDS option since the BOE SDKs allow it.

Other options like BICS and QaaWS only allow live data : the query is run in BW when an end user is consuming the dashboard. When and how many times per user is controlled at the dashboard level.

For non BW data, you will need BOE (with QaaWS or LO) as BICS with Xcelsius only supports BEx queries and views at the moment.

All options support BEx queries, with BICS being the preferred one (all other factors excluded) as LO and QaaWS rely on the OLAP universe and its limitations. See Elements of a BEx query in Xcelsius off BICS for more details on BEx query and the BICS connectivity.

To report off an Infocube you will need an Universe, hence the BOE platform as BICS only supports BEx queries (and views). The Universe can then be accessed from QaaWS, LO or the Webi SDK for the LCDS option.

Only Crystal Reports can access an R3 source at the moment, hence the Live Office - Crystal Reports is the only option there.

Back to comparison table

Deployment and Implementation cost

All options require a BOE platform but the BICS option. For a pure SAP legacy customer with no Business Objects component, BICS is clearly the way to go, unless there is a willingness to invest in the SAP BusinessObjects technology and deploy other products (WebI, Explorer, Crystal Reports, etc). If a BOE environment already exists and the customer staff is trained and comfortable with it, all options can be considered.

From a Life Cycle Management standpoint, BICS leverages transports on the ABAP backend. QaaWS and LO rely on the BOE Life Cycle Management. LCDS requires some custom development and as such weaker than the other options on this point.

Assuming the NetWeaver stack is already in place, here are factors to consider when estimating the implementation cost :

• the need for a BOE environment with all options except BICS
• the need for report creation and maintenance with LO
• the need for some (heavy) custom Java development with the LCDS option, as well as licenses from Adobe for a production LCDS ES2 system

Back to comparison table

Performance comparison

There is no comparison data available at this point from a rigorous benchmark but here are points worth considering.

• data loads in BW are usually done once a day / week, hence a dashboard may not require live access to BW all the time and can leverage the query caching options mentioned below;
• if live data is absolutely required, the BICS option is theoretically performing better than QaaWS or LCDS for 2 reasons :
  • no BOE platform is required, taking a lot of layers off the deployment
  • OLAP universes are using the OLAP BAPI interface which is known to less perform than BICS
• the main selling point of the LCDS option is the native binary format the Xcelsius flash file is fed with. This allows very fast performance when parsing the result set and meant to be used for real time data dashboards. All the other options provide a text XML response which the client web browser has to parse and convert into the native binary format, which is an expensive operation (memory and processing). Sources on the web take about up to 10 times faster; this is the time for the flash file to load the data. The overall performance includes of course the query execution (like for other options) but also heavily depends on your implementation with the Java SDK, whether you implemented some sort of caching, etc.
• regardless of the option you choose, you would benefit from BWA

Query caching is the ability to cache the data retrieved for one filter combination and re-use it when the user selects the same filter combination, avoiding a hit to the backend. This is possible with the LO-CR option, LCDS via custom SDK code and to a certain extent BICS.

• With LO-CR, you schedule a parametrized CR and create the LO document so it uses that parameter to select the filter combination. The CR instance contains all filter combination and is then used as a cache for LO.
• With LCDS, it is the same idea (leverage a scheduled report instance) but you need to implement the logic yourself with custom SDK-based code.
• With BICS, you can leverage the OLAP Cache and load it with the BEx query result set before the users need it. This would typically be done right after the data load with the broadcast feature of Information Broadcasting.

Back to comparison table

References and Links