Lenovo Database Configuration Guide

Transcription

1 Lenovo Database Configuration Guide for Microsoft SQL Server OLTP on ThinkAgile SXM Reduce time to value with validated hardware configurations up to 2 million TPM in a DS14v2 VM SQL Server in an Infrastructure as a Service Model ThinkAgile SXM or SX for Microsoft Azure Stack is an integrated, engineered solution that comes direct from Lenovo ready to go with all features, support, and deployment services included. With benefits like IT agility, lower TCO, and a transformative customer experience, ThinkAgile SXM provides the ease and speed of a public cloud with the security and control of on-premises IT. Azure Stack is an extension of Azure Microsoft s public Cloud offering. Azure Stack is deployed on-premises in a customer s datacenter. As one of the fastest growing transactional database platforms, Microsoft SQL Server deployments are becoming increasingly critical to organizations. They are used in everything from enterprise operational databases to business mission-critical applications, including enterprise online sales transactions, finance management and business intelligence. Microsoft SQL Server is the optimal transactional database for these applications, with breakthrough transactional performance, end-to-end protection, simplified management and support for high availability deployments. This Lenovo Database Configuration guide for Microsoft SQL Server OLTP on ThinkAgile SXM is delivered to market on a high availability hyperconverged platform upon which the Microsoft SQL transactional database is deployed. Lenovo and Microsoft worked together to engineer the platform components Azure Stack software and Lenovo software-defined infrastructure to make sure they perform seamlessly. There is no need to worry about configuring to optimize settings for virtual or physical infrastructure. IT can now focus more on building proficiencies in operating cloud services such as IaaS and SaaS and less on building infrastructure. Scalability and Agility for Transactional Processing Workloads The Lenovo ThinkAgile SXM platform, which incorporates DevOps practices, enables administrators to achieve business agility. This yields an offering enabling rapid deployment of new Microsoft SQL Server instances and VM based transactional databases. Virtual machine sizes offered in Azure Stack are a subset of those offered in Azure. These sizes are listed at the following Microsoft website: Microsoft Azure Stack rules impose limits on virtual machines and resources to avoid noisy neighbor problems where one tenant over consumes resources available on the stamp and starves other tenants. Highlights Offers performance and sizing information for IT organizations looking to implement SQL Server IaaS Delivered as a pre-integrated solution on ThinkAgile SXM that helps to accelerate faster time to deployment and rapid provisioning of the database services Yields a strong price/performance for a SQL transactional database that features built-in high availability Provides a scalable model for transactional processing workloads Enables SQL Server database in Infrastructure as a service (IaaS) implementation model

2 MICROSOFT SQL SERVER 2017 ON THINKAGILE SX FOR MICROSOFT AZURE STACK ThinkAgile SX for Microsoft Azure Stack consists of the following hardware components: From 4 to 12 ThinkSystem SR650 hyperconverged nodes. The SR650 for ThinkAgile SXM requires two Intel Xeon Gold or Platinum processors. The SR650 for ThinkAgile SXM supports 384 GB, 768 GB, or 1,536 GB of memory with 24x 16 GB or 32 GB TruDDR4 RDIMMs or 64 GB TruDDR LRDIMMs. Each SR650 hyperconverged node has 14x 3.5" hot-swap drive bays (12 on the front; 2 on the rear) and an internal M.2 Kit with Mirroring for 2x M.2 non-hotswap SSDs. The 3.5-inch drive bays are used for data storage, and they are connected to the ThinkSystem i SAS/SATA HBA (JBOD mode). For the VM/storage network, the SR650 for ThinkAgile SXM provides two-port 25 GbE connectivity with the dual-port 25 GbE Mellanox ConnectX-4 Lx network adapter The SR630 management node provides hardware management services and includes Windows Server 2016 and Lenovo XClarity The SR650 hyperconverged nodes provide compute and storage resources and include Windows Server 2016 with Hyper-V, Azure Stack software and Storage Spaces Direct (S2D) storage. Network connectivity is provided by the Lenovo switches. The network infrastructure consists of one G GbE switch that runs Enterprise Network Operating System (ENOS), and the VM/storage network consists of two NE GbE switches that run Cloud Network Operating System (CNOS). SQL Server 2017 represents a major step towards making SQL Server a platform that gives you choices of development languages, data types, on-premises or cloud, and operating systems by bringing the power of SQL Server to Linux, Linux-based Docker containers, and Windows. SQL Server 2017 includes many new Database Engine features, enhancements, and performance improvements: Resumable online index rebuild resumes an online index rebuild operation from where it stopped after a failure (such as a failover to a replica or insufficient disk space), or pauses and later resumes an online index rebuild operation. The IDENTITY_CACHE option for ALTER DATABASE SCOPED CONFIGURATION allows you to avoid gaps in the values of identity columns if a server restarts unexpectedly or fails over to a secondary server. A new generation of query processing improvements that will adapt optimization strategies to your application workload s runtime conditions. For this first version of the adaptive query processing feature family, we have three new improvements: batch mode adaptive joins, batch mode memory grant feedback, and interleaved execution for multi-statement table valued functions. Automatic database tuning provides insight into potential query performance problems, recommends solutions, and can automatically fix identified problems. The tested configuration features the following main components: Hardware Server: Lenovo ThinkSystem SR650 (4 nodes) Processors: 2x Intel Xeon Gold GHz Memory: 1536 GB of TRU DDR4 memory Storage: OS Storage: Two 480 GB SATA SSDs for the operating system (RAID 1) Caching tier: 4x 1.6TB 12G SAS SSDs per node Capacity tier: 10x 6TB NL SATA HDDs per node Network: Mellanox 2x 25GbE RDMA NIC 2x Lenovo NE GbE TOR switches 1x Lenovo G8052 1GbE switch for BMC traffic Software: Microsoft Azure Stack build 1802 Microsoft Windows Server 2016 Standard Edition Microsoft SQL Server 2017 Enterprise Edition

3 Test setup and execution Azure Stack offers VM sizes in the General Purpose and Memory optimized categories. The General Purpose VMs range from Basic A series, Standard A series, D-series, DS-series, Dv2-series to DSv2-series. Memory optimized VMs range from D-series, DS-series, Dv2- series to DSv2-series with more memory and more data disks per VM. The memory optimized VMs are more suitable for database functionality. For the purpose of this configuration guide, testing was done with DS12v2, DS13v2 and DS14v2 VMs. Windows Server 2016 and SQL Server 2017 were installed in the VMs. For a balanced and optimized OLTP configuration: Create a DS14v2 VM Only one VM can be created at a time in the Azure Resource Manager (ARM) web session. To create multiple simultaneous VMs, open multiple ARM sessions. After the VM is created, add data disks to host the SQL database and logs For the DS14v2 VM, up to 32 disks are available For maximum I/O performance, use all 32 data disks, with 24 for databases and 8 for logs. Host cache was not enabled on the data disks, tests showed no performance benefit Disk type is HDD, even though it has SSD option, this is not supported yet and the disks will just be HDD backed. Add all of the disks first, then click save to attach them all at once to the VM. Connect to the VM and open Windows Disk Management Create 2 Windows RAID 0 striped volumes as follows: One 24 disk striped volume for databases, 64k allocation size One 8 disk striped volume for logs, 64k allocation size SQL Server settings for optimal performance Set the server s virtual memory to a fixed size of 4096 (not system managed) Set the maximum SQL memory use to 90% of the servers total memory Enable SQL lightweight pooling Enable SQL priority boost Set max worker threads to 3000 Set recovery interval to At this point, the SQL database can be created using HammerDB and then performance tests run. Detailed SQL test setup steps are available in the HammerDB documentation at: HammerDB test parameters used are listed at the end of this document HammerDB can be run on the SQL server or as a remote client Backup the database after it is built, and restore it before each test and restart SQL If run remotely, HammerDB requires the SQL server client installed and Named Pipes enabled on the SQL server, in SQL Configuration Manager, Network protocols. Virtual machines must be created in the same resource group in order to connect, this places them on the same subnet Solution Benefits Performance and sizing for Microsoft SQL Server OLTP workload on Lenovo ThinkAgile SXM offering Features T-shirt sizes for Microsoft SQL Server VMs Integrated System from Lenovo for Microsoft Hybrid Cloud Reduced time to value with pre-tested VMs for Microsoft SQL Server

4

5 HammerDB Configuration, Parameters and Performance

6 Lenovo ThinkAgile SX for Microsoft Azure Stack yields a hyper-converged highly available, high performance platform for an OLTP solution Lenovo ThinkAgile SX for Microsoft Azure Stack ThinkAgile is a preintegrated, engineered solution that comes direct from Lenovo ready to go with all features, support, and deployment services included. With benefits like IT agility, lower TCO, and a transformative customer experience, ThinkAgile provides the ease and speed of a public cloud with the security and control of on premises IT. Microsoft Azure Stack provides a subset of Azure cloud services. The Azure Stack software is deployed on powerful, ultra reliable, and innovative Lenovo ThinkSystem SR650 nodes in a hyperconverged, clustered configuration engineered for high stability, resilience, and performance. Azure Stack enables consistent cloud services across the Microsoft Azure environment. The internal networking for this solution is 25GbE, provided by the cloud-optimized Lenovo ThinkSystem NE2572. This enterprise-class Layer 2 and Layer 3 full featured switch delivers line-rate, highbandwidth switching, filtering, traffic queuing and high-availability features without delaying data. Large data center-grade buffers help keep traffic moving, while the hot-swap redundant power supplies and fans help provide resiliency for business-sensitive traffic. External connectivity from the ThinkSystem NE2572 is afforded by 6 QSFP 100GbE ports. Why Lenovo Lenovo is a leading provider of x86 servers for the data center. Featuring rack, tower, blade, dense and converged systems, the Lenovo server portfolio provides excellent performance, reliability and security. Lenovo also offers a full range of networking, storage, software, solutions, and comprehensive services supporting business needs throughout the IT lifecycle. With options for planning, deployment, and support, Lenovo offers expertise and services needed to deliver better service-level agreements and generate greater end-user satisfaction. Lenovo ThinkAgile SX for Microsoft Azure Stack For More Information To learn more about Lenovo Database Configuration for Microsoft SQL Server OLTP on ThinkAgile SXM, contact your Lenovo Business Partner or visit: Lenovo. All rights reserved. Availability: Offers, prices, specifications and availability may change without notice. Lenovo is not responsible for photographic or typographical errors. Warranty: For a copy of applicable warranties, write to: Lenovo Warranty Information, 1009 Think Place, Morrisville, NC, 27560, Lenovo makes no representation or warranty regarding third party products or services. Trademarks: Lenovo, the Lenovo logo, System x, ThinkServer are trademarks or registered trademarks of Lenovo. Microsoft and Windows are registered trademarks of Microsoft Corporation. Intel, the Intel logo, Xeon and Xeon Inside are registered trademarks of Intel Corporation in the U.S. and other countries. Other company, product, and service names may be trademarks or service marks of others. CRN: DBSMS02XX81 04/2018