Technical Paper Hardware Guidelines for a 16-Server, 256-Core SAS® Visual Analytics 7.1 Environment Prepared by the Enterprise Excellence Center Release Information Content Version: 1.0 December 2014. Trademarks and Patents SAS Institute Inc., SAS Campus Drive, Cary, North Carolina 27513. SAS® and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are registered trademarks or trademarks of their respective companies. Statement of Usage This document is provided for informational purposes. This document might contain approaches, techniques, and other information proprietary to SAS. Overview SAS® Visual Analytics is a highly scalable solution that allows users to add servers as needed to achieve their objectives. Each environment provides users with a discrete amount of resources that might have various performance attributes depending on the work. These guidelines specifically outline the configuration and performance characteristics for a 16server, 256-core environment. The objective of this environment is to provide excellent response times for the majority of single-function requests. Users can review these guidelines to evaluate the feasibility of the 16-server, 256-core environment for their use. This configuration represents the minimum standard distributed environment. An official sizing is recommended to ensure performance expectations. Work with your account executive to have an official sizing performed. Configuration Specifics and Requirements SAS Visual Analytics 7.1 requires SAS 9.4. o Alternative SAS deployments are not required to be upgraded. Intel Xeon Haswell E5-2667 v3 dual 8-core processors (16 cores total) were used. o Alternative processors will work. Results will vary and an official sizing is recommended. 256 GB of RAM using LRDIMMs ((16) 16 GB 2133 MHz LRDIMMs per node). Red Hat Enterprise Linux 6.4 or higher. Distributed SAS Visual Analytics configurations can be virtualized. Usage Considerations Here are three key considerations for evaluating the appropriateness of the environment: 1. Identify the size of the largest data table to be loaded into RAM. 2. Identify the total amount of data to be loaded into RAM and the required disk storage. 3. Quantify the total number of concurrent users, broken down into heavy and light users. By our definition, a heavy user is potentially any SAS Visual Analytics Explorer user or a user who runs correlational analysis with multiple variables, box plots with four or more measures, or crosstabs with four or more class variables. As the size of the data table increases, response times become longer. A light user runs simple line charts, box plots, bubble plots, or geo maps with one or two class variables and up to five measures or a user who runs only reports. Performance Considerations The following chart shows the relationship between the number of users and the table size. The maximum number of users is on the Y axis and the maximum data table size in gigabytes is on the X axis. Adding more users than the chart displays can result in performance degradation. Using larger data tables impacts the response times and the overall number of concurrent users. 1. Find the size of your largest table in the chart below. If your largest table is larger than 400 GB, this environment is too small if you are looking for 15-seconds-or-less response times. 2. Read the chart to determine the maximum number of users for a table size. If the number of users exceeds the maximum identified in the chart, this environment is too small. 3. Be advised that different chips produce different results because their processing speeds are different. An official sizing is always recommended. This example uses the Intel Xeon E5-2667 v3 dual 8-core 3.2 GHz chip. 4. The ratio of heavy users to light users is 1 to 3. For every heavy user that you subtract, you can add 3 light users. For example, a 200 GB data table has a maximum of 11 heavy users. If you have only 4 heavy users, take the remaining 7 and multiply by 3. The result is a total number of 4 heavy users and 21 light users for the 200 GB data table on this distributed SAS Visual Analytics configuration. Hardware Configuration These guidelines are based on commodity hardware from Dell using the Intel Xeon architecture. All major hardware vendors can provide commodity machines that meet the hardware specifications. SAS Visual Analytics 16-Server, 256-Core Distributed Environment Number of servers (16) HP ProLiant DL360 G9 Processors per server (2) 3.2 GHz 8-Core Intel Xeon E5-2667 v3 Total cores 256 RAM per server 256 GB ((16) 16 GB 2133 MHz LRDIMMs) Disk per server 2 x 600 GB (10K RPM) Usable RAM ~2.6 TB Usable Disk ~6.5 TB Operating System Red Hat Enterprise Linux 6.4 or higher NIC Dual 10 GB NIC Adapter Card * Server power settings need to be set to the maximum, not to the factory setting. * Hyper-threading is recommended for all production CPUs. * Some features of SAS Visual Analytics 7.1 (such as text analytics and compression) impact system performance. This sizing does not include that impact. * This sizing assumes that all data is uncompressed. Critical Data Considerations These guidelines do not address the data management resources needed outside of SAS Visual Analytics. Getting data into SAS Visual Analytics and performing other ETL functions are solely the responsibility of the user. The maximum single largest table size is 400 GB. Larger tables can be used, but performance starts to degrade (for example, slow down). SAS Visual Analytics can run with a single data set or multiple data sets loaded into memory. Up to ~2.6 TB of data tables can be loaded into memory. The Enterprise Excellence Center Sizing team can provide customer-specific configurations for detailed planned deployments based on the customer’s proffered hardware vendor. Any changes to the hardware baseline or data specifications invalidates these guidelines. Frequently Asked Questions What will happen if we increase the number of users or the amount of data for analysis to our system? As the use of SAS Visual Analytics grows throughout your organization, it is common to add new users and provide more data for analysis. Depending on the number and type of increase, it is reasonable to expect that when the number of concurrent sessions or data tables to be accessed and available in memory increases, the performance of the system will degrade. The system is easily scalable. Additional servers of the same configuration can be added to provide the resources needed to scale based on your business goals. Work with your account executive to have an official sizing performed. What if our standard is a different Intel Xeon processor or processor family? The SAS Visual Analytics distributed environment is a supported commodity 2- or 4-core machine with the Intel Xeon E52600 v2 or v3 series processor with a clock speed of at least 2.6 GHz and a RAM configuration that provides a minimum of 1600 MHz memory speed. The combination of the amount of data needed, the amount of data loaded into memory, and the number and type of concurrent user sessions can have significant performance implications. An official sizing is recommended to ensure successful implementation. The number of cores per server can have an impact on licensing. Work with your account executive to have an official sizing performed. Why not just add more RAM? In some configurations, you might be able to increase the RAM per server to increase the available usable memory. In other configurations, the increase in RAM can cause the memory speed to dramatically drop (sometimes as much as 50%). This can negatively impact the overall performance of the environment. An official sizing is recommended to ensure performance expectations. Work with your account executive to have an official sizing performed. What other options are available for the 16-server distributed SAS Visual Analytics environment? For the configuration, there are multiple chip, memory, and disk options depending on the make and model of the server being used. All of these options can have an impact on the maximum data table size and maximum number of users. These guidelines are just one example and are recommended as a starting point. To get more exact information, an official sizing is required. To contact your local SAS office, please visit: sas.com/offices SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc. in the USA and other countries. ® indicates USA registration. Other brand and product names are trademarks of their respective companies. Copyright © 2014, SAS Institute Inc. All rights reserved.
© Copyright 2024