Deploying Mysql Databases On Red Hat Ceph Storage-PDF Download

  • Date:01 Jul 2020
  • Views:3
  • Downloads:0
  • Pages:24
  • Size:1.32 MB

Share Pdf : Deploying Mysql Databases On Red Hat Ceph Storage

Download and Preview : Deploying Mysql Databases On Red Hat Ceph Storage

Report CopyRight/DMCA Form For : Deploying Mysql Databases On Red Hat Ceph Storage


6 RED HAT TEST ENVIRONMENT 17,Lab configuration 17. Sysbench 19,7 TUNING AND BEST PRACTICES 20,Optimizing MySQL for Ceph 20. Opitimizing Ceph 20, Deploying on Linux containers versus virtual machines 22. 8 SUMMARY 23, redhat com TECHNOLOGY DETAIL Deploying MySQL Databases on Red Hat Ceph Storage 2. INTRODUCTION, Both legacy and cloud native applications are benefiting from design patterns that emerged from.
the public cloud Diverse organizations are now looking to model the very successful public cloud. Database as a Service DaaS experiences with private or hybrid clouds using their own hardware. and resources The combination of OpenStack and Red Hat Ceph Storage can provide an environ. ment that is familiar to anyone who has built applications atop services like Amazon Elastic Compute. Cloud and Elastic Block Store EBS or Google Compute Engine and Persistent Disk Providing a. high level of fidelity with public cloud gives developers and operators alike a familiar environ. ment in which to work increasing application agility all while offering better price performance. characteristics, OpenStack cloud deployments are growing and organizations are increasingly selecting both MySQL. databases and Ceph storage in their OpenStack environments The sheer number of databases is. growing dramatically as well with many database administrators DBAs reporting that they manage. hundreds to thousands of separate databases While virtualized server environments have helped to. enable this growth inflexible siloed storage infrastructure has made limited progress In fact rigid. and monolithic storage infrastructure is often an impediment to either effective database scalability. or effective management of multi database environments. As an open software defined storage platform Red Hat Ceph Storage provides a compelling solution. Ceph storage clusters can serve diverse types of workloads with carefully chosen and configured. hardware Though throughput intensive and cost capacity focused workloads are common using. Ceph to serve IOPS intensive MySQL workloads is is increasingly seen in production environments. Cloud like MySQL storage solutions are made possible by the availability of flash storage media. high performance MySQL implementations from Percona and Ceph optimized hardware platforms. from companies like Supermicro, Any storage solution for MySQL must provide sufficient low latency IOPS throughput to support the. needs of key databases and applications at a cost point that is comparable to public cloud infra. structure Working closely with Percona and Supermicro Red Hat has conducted extensive evalua. tion and testing of MySQL workloads on Red Hat Ceph Storage clusters Results clearly demonstrate. that Percona Server for MySQL running on Red Hat Ceph Storage and Supermicro servers compares. favorably with common public cloud solutions in terms of both cost and performance Specifically. organizations that have chosen a private or hybrid cloud model can provide SSD backed Ceph. RADOS Block Device RBD storage at a price point that is even more favorable than public cloud. offerings while retaining essential performance characteristics. redhat com TECHNOLOGY DETAIL Deploying MySQL Databases on Red Hat Ceph Storage 3. DEPLOYING MYSQL ON CEPH CLUSTERS, Modern DBAs have many deployment options for MySQL and similar database technologies To. evaluate Ceph storage solutions it is important to understand both the technical synergies and the. trends that are shaping MySQL and OpenStack environments. CEPH AND MYSQL A COMPELLING TECHNOLOGY COMBINATION. Many organizations have become comfortable with deploying MySQL for their applications in the. public cloud As a result storage technology for hybrid or private cloud MySQL deployments should. emulate public cloud methods Ceph and MySQL represent highly complementary technologies. Strong synergies MySQL OpenStack and Ceph are often chosen to work together Ceph is the. leading open source software defined storage solution MySQL is the leading open source rela. tional database management system RDBMS 1 Moreover Ceph is the number one block storage. for OpenStack clouds2 with MySQL based applications figuring as the number four OpenStack. Operational efficiencies Ceph storage contributes directly to operational efficiencies for MySQL. databases Ceph provides a shared elastic storage pool with flexible volume resizing and dynamic. database placement Live instance migration is supported With Ceph operators can back up to an. object pool and read replicas are easily created via copy on write snapshots. Public cloud fidelity Developers want platform consistency and effective private or hybrid clouds. require familiar patterns to those established by existing public clouds Ceph provides block and. object storage like public cloud solutions with consistent storage features Table 1 while letting. organizations use their own hardware data centers and staff. TABLE 1 COMPARING CEPH WITH PUBLIC CLOUD STORAGE SOLUTIONS. FEATURE CEPH RBD GOOGLE AMAZON ELASTIC,PERSISTENT DISK BLOCK STORE EBS. FAULT TOLERANCE Yes Yes Yes,SNAPSHOTS Yes Yes Yes,VOLUME RESIZING Live Live Detached.
VOLUME MIGRATION Yes Yes Yes,VOLUME LIVE Yes Yes No. ZONE MIGRATION Detached Live Detached,READ CHECKSUMS No Yes No. OPENSTACK DATABASE AND STORAGE TRENDS, OpenStack has been adopted by organizations of all shapes and sizes to provide cloud infrastructure. services for internal developers and operators Results of the OpenStack user survey are published. at their bi annual OpenStack Summit Starting with the November 2014 survey and continuing in the. most recent survey Ceph RBD has dominated the polls as the top Cinder driver for providing block. storage services in both production and development environments. 1 gartner com doc 3033819 state opensource rdbmss, 2 Ceph is and has been the leading storage for OpenStack according to several semi annual OpenStack user surveys. redhat com TECHNOLOGY DETAIL Deploying MySQL Databases on Red Hat Ceph Storage 4. Figure 1 contrasts the number of survey respondents choosing Ceph RBD in production environ. ments compared to the second most common Cinder driver Logical Volume Management LVM. This trend is particularly impressive given that LVM is the default reference driver At the April 2016. OpenStack Summit the Cinder development team announced that Ceph RBD would be elevated to. the status of reference driver alongside LVM,Platform usage for Cinder.
November 2014 May 2015 October 2015 April 2016, Figure 1 Ceph continues to dominate polls as the number one Cinder driver for production block storage services. The OpenStack User Survey also collected information about which application stacks were the most. popular in production OpenStack environments Figure 2 shows the percentage of respondents that. reported using a particular application framework in their OpenStack environment for the October. 2015 and April 2016 surveys The results illustrate the popularity of MySQL for both custom and off. the shelf applications running on OpenStack clouds. Ruby on Rails,Windows ISS SQL server ASP WISA,Linux Apache MySQL and PHP LAMP. 0 10 20 30 40 50 60 70 80,October 2015 April 2016,Figure 2 OpenStack application framework trends. redhat com TECHNOLOGY DETAIL Deploying MySQL Databases on Red Hat Ceph Storage 5. Recent surveys have asked respondents which database they use to underpin the OpenStack. control plane MySQL derived databases dominate the standings Figure 3 The results also show. an increased adoption of Galera clusters with either MySQL or MariaDB This is likely a sign that. OpenStack environments are adopting business critical applications making a highly available. control plane increasingly attractive The chart below depicts OpenStack database trends. Percentages,MySQL Percona MariaDB with,MySQL MariaDB MangoDB Percona Server Postgres. with Galera XtraDB Cluster Galera,November 2014 October 2015 April 2016.
Figure 3 MySQL continues to lead Openstack database trends. The results of these surveys echo conversations with Red Hat customers The common thread is that. application migration to OpenStack environments is accelerating and that databases are a critical. component of many of these applications As more applications are being migrated the demand. for IOPS optimized storage volumes is growing Simply our customers have been asking how to. deliver IOPS optimized storage volumes using their favorite open source software defined storage. technology Ceph,SUPPLEMENTING OTHER ARCHITECTURAL CHOICES. DBAs have many architectural options available to them when designing database services to. support an organization s applications A number of objectives can influence the deployment archi. tecture including database size schema design table size number of databases availability write. and read scalability and consistency model Table 2 and the sections that follow provide a summary. of relevant architectural choices and objectives along with implications for Ceph storage. redhat com TECHNOLOGY DETAIL Deploying MySQL Databases on Red Hat Ceph Storage 6. TABLE 2 FACTORS DRIVING MYSQL ARCHITECTURE DECISIONS. MYSQL MYSQL MYSQL,FACTORS ON CLUSTER GALERA MYSQL ON MYSQL. CEPH NBD CLUSTER REPLICATION DRBD SHARDING,LARGER THAN Yes Yes No No No Yes. SINGLE HOST,TABLE LARGER,THAN SINGLE Yes Yes No No No No. ON DATASET Yes No No No No No,PARALLELISM,TOLERANT Yes Yes Yes Yes Yes No.
READ SCALING N nodes 48 nodes Yes Yes No N nodes, WRITE N nodes 24 nodes 1 node 1 node 1 node N nodes. CONSISTENT Synch Synch Synch Synch,WRITES Asynchronous N A. ronous ronous ronous ronous,MULTI No Yes Yes No No No. MySQL on Ceph RBD, When managing a large collection of MySQL databases across a large number of servers database. operators must balance the load of the collection by consolidating databases onto servers with other. databases that each present unique resource constraints Storage is typically a hard constraint due. to its spatial nature A target server either does or does not have sufficient remaining space When. databases have a storage footprint approaching the size of a single server but without commensu. rate demands on the rest of the server s resources those other resources become trapped capacity. Lower utilization is the result, A large deployment of servers represents a significant financial investment both in capital and.
operational expenditure Recapturing trapped capacity is one of the major drivers for implementing. distributed storage technologies like Ceph Ceph lets groups of servers pool their storage resources. and allows any server in the group to access storage over the network This in turn allows more flex. ible and efficient placement of databases across the cluster without requiring tools like rsync or. MySQL replication to move data Ceph also provides a wide range of features including live volume. resize volume mirroring volume migration virtual machine live migration copy on write cloning. and snapshots both full and differential Ceph is thus an ideal fit for managing large collections of. MySQL database instances, redhat com TECHNOLOGY DETAIL Deploying MySQL Databases on Red Hat Ceph Storage 7. MySQL Cluster NDB, MySQL Cluster is a multi master MySQL database technology that is shared nothing with no single. point of failure There are three types of nodes processes in a MySQL Cluster. Management nodes are responsible for the configuration monitoring and starting or stopping. the cluster, Application nodes provide a familiar SQL interface and execute queries on behalf of application. Data nodes store the actual data of the cluster and are grouped into pairs that each manage par. titions or shards and transparently handle replication self healing and load sharing. MySQL Cluster was originally designed for the telecommunications market to store subscriber data. where high write performance and high availability were paramount MySQL Cluster shards at the. application level by hashing a table s primary key or hashing a hidden primary key if a primary. key is absent from a table This ability lets a single table grow well beyond the storage capacity of. a single host because a table can be partitioned into up to 192 shards or eight partitions per node. group The entire MySQL cluster can support up to 24 node groups or 48 data nodes Since there is. no straightforward way to identify which database is consuming CPU resources MySQL Cluster is. relegated to single tenant applications,Galera Cluster for MySQL. Galera Cluster for MySQL provides nearly synchronous multi master replication is highly available. and provides read scalability without slave lag or error prone master failovers A Galera Cluster is. known as a primary component in Galera parlance Galera Cluster does not provide its own load. balancing component so it is often deployed in conjunction with load balancing software such as. HAproxy to spread load across members of the cluster All data is near synchronously written to. every member of the cluster Having data on every node means write performance is bound by the. round trip time to the farthest node Galera Cluster s replication strategy constrains the size of a. database to the capacity of the smallest node, Galera Cluster for MySQL can use Ceph block devices to avoid storage constraints and allow rapid.
reconstitution of a failed member by moving the volume to a new node Ceph snapshots also allow. Both legacy and cloud native applications are benefiting from design patterns that emerged from the public cloud Diverse organizations are now looking to model the very successful public cloud Database as a Service DaaS experiences with private or hybrid clouds using their own hardware and resources The combination of OpenStack and Red Hat Ceph Storage can provide an environ ment that

Related Books