Compatability requirements and sizing your BusinessObjects landscape Dr. Berg Comerit Inc. © 2011 Wellesley Information Services. All rights reserved. This Session • This session explores important system sizing concepts for performance • We look a five real-world sizing examples • We will analyze some of the factors that could influence your sizing and scaling decisions. • Look at in-memory processing options with HANA and BWA • You will get details how to do a sizing for your dashboards and software configuration • Take-home resource: a T-Shirt sizing model you can adapt to your requirements 1 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 2 The Components of BI 4.0 The complete components of the BI landscape consists of a variety of software. Each of these components have their own requirements Many customers will connect their dashboards directly and will not use BO 4.0 as their deployment method 3 The NetWeaver Version Requirements for Analysis BICS connections There are several version possibilities to take advantage of the BICS connectors for OLAP - Analysis. Make sure your BW system is at a minimum level. Before you start, make sure your SAP BW system is at least on one of these options: BW 7.0 ServicePack 23 BW 7.0 Ehpk 1, ServicePack 5 BW 7.3 , Any ServicePack BICS connectors are 20-40% faster than MDX based OLAP Universes, so this interface is what you really want to make your BI 4.0 really fast 4 Hardware BI 4.0 Server Side Requirements From a Server sizing perspective you need: Minimum CPU POWER5 (IBM-AIX) UltraSPRACT24 (Sun) 2.0 GHz Intel Core CPU (Windows & Linux); Compiled files using Windows, can use CPUs from both Intel and AMD. Minimum Memory of Server Min of 4.0 GB Memory (more based on number of users) Minimum Disk Space: If you only install English: 11 GB Windows; 13 GB AIX/Solaris; and 14 GB for Linux If you install all languages: 14 GB Windows; 15 GB AIX/Solaris; and 16 GB for Linux This is where sizing the environment really matters. We will come back with some sizing guidelines later, but these are minimum requirements 5 Operating System BI 4.0 Server Side Requirements From an OS on the Server side, the following versions are supported All server components only support 64-bit operating systems The Mobile Server is currently only supported on Windows operating systems 6 Data Base Server Options Microsoft SQL Server Express 2008 is included with BI 4.0, but you may also use databases from Oracle, Sybase or IBM. The database is used for storing CMS and the audit repository of BI. Using the SQL Server database that comes bundled with BI 4.0 may save you a lot of money 7 Application Server Requirements The Application server handles the logic of BI 4.0 and can run on a variety of hardware. The core supported application servers include: Note: AIX supports WebLogic Servers with Java 5. Sizing the application server is part of the BI 4.0 sizing and has changed to be based on SAPS instead of CPU and Memory (as we did in XI 3.x). 8 Web Portal Requirements - BI 4.0 SP1 The web rendering occurs through a web portal. You can use WebLogic, WebSphere or the standard SAP Enterprise Portal All major operating systems are supported for the web portal Source: SDN: SF. Gonzales, 2011 There are many severs and components to consider. Sizing each component is critical to overall solid performance. 9 PC Hardware Requirements Client Side BI 4.0 From a PC client perspective you need: Memory Min of 1.0 GB Memory (really want more in practice) I recommend: 2.0 GB (or more if you can afford it) Disk Space: Min of 3.5 GB free space if you only install English Min of 7.5 GB free space if you install all languages Screen Size Recommended resolution size is Make sure you build Dashboards on a standardized screen resolution and size so that everyone sees the same images 10 Operating System - Client Side Requirements BI 4.0 Most Windows Operating systems are supported by BI 4.0 at the client side. While all desktop components of BI 4.0 are 32-bit, you can use either 32-bit of 64 bit operating systems Mac OS/X is currently not supported for other than the Crystal report viewer 11 Web Browser Requirements There are several different browser requirements for WebIntelligence and Analysis, this list is specifically for dashbords. Please consult other products for their browser requirements. Quick rule: when in doubt, go for the higher versions (IE 8 or Firefox 3.6). Also only 32bit browsers are supported. The Apple Safari Browser 4.0 is supported only by the Mac OS X operating system 12 Other Software Requirements To run the system correctly, there are several components needed on the client side. The use of these depends on the software deployed. These is required: To use the enterprise monitoring feature of BI 4.0, you need Flash version 10. SAP will no longer support Windows XP in the next 4.1 release. Office 2010 is now also Supported!! Some have reported issues with large flash files. For example: Flash v10.1 seems to be stuck in 'initializing' phase when files are over 2MB. This seems to less of an issue for smaller files and many have reported no significant issues in version 10 13 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 14 Real-Word Examples Since most client has only started to test BI 4.0, so the numbers below are for planned rollouts for five companies that have installed BI 4.0 on their hardware platforms and have a go-live this fall. Performance numbers are based on initial testing and limited rollouts. Portal Dashboards Analysis WebI Hardware BW Version Named Users (#) Concurrent Users (#) Simultaneous Requests (#) Named Users (#) Concurrent Users (#) Simultaneous Requests (#) Named Users (#) Concurrent Users (#) Simultaneous Requests (#) Server Memory Server Disk PC Memory (standard) PC CPUs (standard) Other Portal version Server Operating System Flash version Database Version Performance overall (1-10) Manufacturing Company 7.0 Enpk 1 ~175 5-8 5-10 40 6 4-5 8 GB 100 GB 2 GB 2.33 GHz (dual core) WebSphere AIX 9 SQL express 9 Airline 7.0 Enpk 1 ~30 6-10 4-10 84 22 5-15 4 GB 75 GB 2 GB 2.0 GHz SAP v7 Win 2008 10 SQL express 8 Pharma distributor Paper company Defense company 7.0 7.0 Enpk 1 7.0 Enpk 1 168 42 26 11 4-20 3-8 ~30 ~100 7 14-15 2-3 7-10 ~30 ~1800 7 ~80-100 2-3 20-50 8 GB 8 GB 16 GB 100 GB 50 GB 150 GB 2 GB 1 GB 2-4 GB 2.0 GHz 1.8 GHz (single 2.0 GHz (dual core) core) (dual/quad core) SAP v7 SAP v7 SAP v7 Win 2008 Win 2008 Win 2008 10 9 10 SQL express SQL SQL express 9 7 8 15 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 16 Key Factors that Determines Performance of your Dashboard • Some of the servers: CMS Repository database (minor impact to sizing) File Repository Servers (FRS) Adaptive Processing Server (APS) Page server (CR) and Cache servers (Web Intelligence Processing Servers & other tools) Examples are including BI 4.x server Source: SAP AG 2011 More Key Factors that Determines Dashboard Performance • Concurrent number of users during peak load times of system • Logical design of dashboards Simple, complex, and incredible complex Number of records retrieved by the dashboards • Number of nodes in the cluster (BI 4.0). This is uses for spreading service loads on multiple nodes. • Number of CPUs and Available Memory of each server 18 The SAP BI Scalability BI is highly scalable. If you have a high number of users, you can split the application, processing and intelligence layers on separate hardware severs. You can also horizontally partition the processing and intelligence layers on several servers. Tier Components for Intelligence Tier Components for Processing If you split the web application components on multiple servers, make sure you also deploy a hardware load balancer 19 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 20 The T-Shirt Sizing Model • Calculating the BI 4.0 tier memory is somewhat more complicated. We recommend 4GB per core as the BI 4.0 platform is native 64 bit. Using the information above, 36 GB is the recommended amount. The memory for the Backend tier is largely dependent on the size of the EDW. You should involve your DBA in calculating the Backend memory. PS! Many customers will connect their dashbaords directly and will therefore not use BO 4.0 as their deployment method (they can still link to them in BI 4.0 through a web page). 21 Concurrent Vs. Named Users • SAP has new terminology in their sizing: Power Users is called "expert user" Casual User are called "information consumers" Executives are still called "executive user". • The load on the system depends on the user type. I.e., an information consumer user have typically 20% the number of reloads and navigations of an Expert User (trips back to the backend server) • The system sizing depends on the number of concurrent users. For load testing and sizing we assume a max load of 20% of named users For stress testing we may test at 40% load of the named users. Many do this to see how the system reacts, but do not actually size their system to handle this (cost issue) I recommend using a 40% sizing number if the dashboards are externally facing and poor performance may reflect negatively on the company image and service levels 22 The T-Shirt Sizing Model • • • SAPS is a 'unified' measure of system performance. It allows hardware vendors to benchmark what is needed to support a SAP system. All hardware vendors are familiar with this benchmark. Concurrent Users Named Users** Front-End Backend-End Front-End Backend-End (SAPS) (SAPS) (Memory - GB) (Memory GB) Small (S) 25 125 3,620 1,690 8 8 Medium (M) 50 250 7,230 3,380 16 8 Large (L) 100 500 14,460 5,070 24 16 Extra Large (XL) 150 750 21,690 8,440 36 16 Size * For systems with more than 150 concurrent users, a dedicated architecture should be built ** Assumes 20% concurrent users Source: Jason DeMelo, SAP AG SAP references 'named users' between 10-20%, so a small system (s) can handle between 125 to 250 users depending on load. I recommend using the lower number. 23 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 24 Why In-memory processing? • Disk speed is growing slower than all other hardware components Technology Drivers 1990 2011 CPU 0.05 291.17 MIPS/$ MIPS/$ Memory 0.02 70.15 MB/$ MB/$ 216 264 Addressable Memory Network Speed Disk Data Transfer 100 100 Mbps Gbps 5 702 MBPS MBPS Architectural Drivers Improvement 5823x 3508x 248x 1000 x 1990 2011 Disk-based data storage In-memory data stores Simple consumption of apps (fat client UI, EDI) Multi-channel UI, high event volume, cross industry value chains Generalpurpose, applicationagnostic database Applicationaware and intelligent data management Source: 1990 numbers SAP AG 2011 numbers, Dr. Berg 140x Physical hard drive speeds only grew by 140 times 25 since 1990. All other hardware components grew faster. In Memory Processing - General Highlights - BWA User expectations have changed in the last 2-3 years. Today, dashboard users tends to spend less than 14-20 seconds before clicking the back button, refresh, or abandoning the site. 3. Queries are routed to BWA by the Analytical engine SAP BW BI Analytical Engine 2. Indexes copied in to RAM on blades InfoCubes In-memory processing of the data fetches can significantly improve the performance. 1. Indexing and compression stored on a file system DSOs In Memory Processing - Creating Indexes from external data If you need to get external data into your dashboards, it is tempting to simply load it into a spreadsheet. However, this create significantly more manual steps to maintain the solution. A better way may be to include Data Services in the landscape and move the data directly into memory and query the data from there. HANA - How is it Evolving? Hana is available as version 1.0 and can only replicate data from ECC to an in-memory platform. However, in ServicePack 3, HANA will be able to do the same for BW data. This was formerly known as HANA v 1.5). The long-term idea with HANA is to replace the databases under BW and ECC with in-memory processing databases, instead of traditional relational databases. This means much faster query response time and more compressed data (smaller database). But for now, BWA is for BI data and HANA is for in-memory operational reporting. Vs. 28 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 29 High volume User Management and Access Control • • Plan for a gradual rollout to a limited number of users Keep the numbers comparable if possible. This will allow you to predict system loads and performance issues by stipulations from real performance data. • Simplified versions of high impact dashboards may be created for casual users • i.e. roll out to 50 users each week i.e. a dashboard with only one query and summarized data with limited navigation and passing of variables Create a hardware contingency plan and budget accordingly Only in rare cases should you use a big-bang approach. Since user patterns are hard to predict, this may cause significant performance issues. 30 What We’ll Cover … • • • • • • • • Background The Components of the Landscape Real-Word Examples The SAP Xcelsisus Sizing Components The T-Shirt Sizing Model In-Memory Performance Options High volume User Management and Access Control Wrap-up 31 Current Resources • Creating Dashboards with Xcelsius: Practical Guide by Evan Delodder and Ray Li (SAP PRESS) • SAP BusinessObjects Dashboards 4.0 Cookbook by David Lai and Xavier Hacking (Packt Publishing) • Xelcisus Forum on SDN • http://forums.sdn.sap.com/forum.jspa?forumID=302 Xelcisus Tutorials on SDN http://www.sdn.sap.com/irj/boc/dashboards-elearning • SAP MarketPlace for Sizing guidelines • BusinessObjects Expert (on-line magazine and web site) businessobjectsexpert.wispubs.com 32 7 Key Points to Take Home • There are many areas to performance tune - look at all • Size your system based on concurrent users and SAPS • Use the SAP system guides on marketplace, but plan to operate your system at max. 70% load for 'spare capacity' • Keep the BI 4.0 environment on a separate stack from BW • Give the system its own Java install (do not share with other components) • Make sure the PCs have enough memory • Examine the 'standard' PC of the users and developers; pay attention to connectivity, screen size and resolutions, CPUs and all software release versions to assure compatibility. 33 Your Turn! How to contact me: Dr. Berg bberg@comerit.com Disclaimer SAP, R/3, mySAP, mySAP.com, SAP NetWeaver®, Duet®, PartnerEdge, and other SAP products and services mentioned herein as well as their respective logos are trademarks or registered trademarks of SAP AG in Germany and in several other countries all over the world. All other product and service names mentioned are the trademarks of their respective companies. Wellesley Information Services is neither owned nor controlled by SAP. 35