Severalnines Self-Training: MySQL® Cluster - Part V
•
6 likes•8,971 views
Part V of our free self-training slides on MySQL Cluster. In this part we cover ’Internals' * NDB Design Internals
Report
Share
Recommended for you
Recommended for you
Recommended for you
Recommended for you
Recommended for you
Recommended for you
Recommended for you
Recommended for you
More Related Content
What's hot
What's hot (20)
Similar to Severalnines Self-Training: MySQL® Cluster - Part V
Similar to Severalnines Self-Training: MySQL® Cluster - Part V (20)
More from Severalnines
More from Severalnines (20)
Recently uploaded
Recently uploaded (20)
Severalnines Self-Training: MySQL® Cluster - Part V
- 1. MySQL Cluster Training presented by severalnines.com Address: Contact: SeveralninesAB Jean-Jérôme Schmidt c/o SICS, Box 1263 Email: services@severalnines.com Isafjordsgatan22 SE-164-29 Kista Copyright 2011 Severalnines AB Control your database infrastructure 1
- 2. Introduction • At Severalnines, we believe in sharing information and knowledge; we all come from an open source background • We know a lot of things about MySQL Cluster and think that MySQL Cluster is a great technology • These free MySQL Cluster Training slides are a contribution of ours to the knowledge and information sharing that’s common practice in the open source community • If you have any questions on these slides or would like to book an actual training class, please contact us at: services@severalnines.com Copyright 2011 Severalnines AB Control your database infrastructure 2
- 3. Training Slides - Concept • Over the coming weeks we will be chronologically releasing slides for the different sections of our MySQL Cluster Training program on our website. • The full agenda of the training with all of its modules is outlined in the next slides so that you can see what topics will be covered over the coming weeks. • Particularly specialised topics such as Cluster/J or NDB API are not fully covered in the slides. We recommend our instructor-led training classes for such topics. • Please contact us for more details: services@severalnines.com Copyright 2011 Severalnines AB Control your database infrastructure 3
- 4. Full Training Agenda (1/4) • MySQL Cluster Introduction – MySQL eco system – Scale up, scale out, and sharding – MySQL Cluster Architecture – Use cases – Features – Node types and Roles • Detailed Concepts – Data Distribution – Verifying data distribution – Access Methods – Partitioning – Node failures and failure detection – Network Partitioning – Transactions and Locking – Consistency Model – Redo logging and Checkpointing • Internals – NDB Design Internals Copyright 2011 Severalnines AB Control your database infrastructure 4
- 5. Agenda (2/4) • Installing MySQL Cluster – Setting up MySQL Cluster – Starting/stopping nodes – Recovery and restarts – Upgrading configuration – Upgrading Cluster • Performance Tuning (instructor-led only; contact us at services@severalnines.com) – Differences compared to Innodb/MyISAM – Designing efficient and fast applications – Identifying bottlenecks – Tweaking configuration (OS and MySQL Cluster) – Query Tuning – Schema Design – Index Tuning Copyright 2011 Severalnines AB Control your database infrastructure 5
- 6. Agenda (3/4) • Management and Administration – Backup and Restore – Geographical Replication – Online and offline operations – Ndbinfo tables – Reporting – Single user mode – Scaling Cluster • Disk Data – Use cases – Limitations – Best practice configuration • Designing a Cluster – Capacity Planning and Dimensioning – Hardware recommendations – Best practice Configuration – Storage calculations Copyright 2011 Severalnines AB Control your database infrastructure 6
- 7. Agenda (4/4) • Resolving Issues – Common problems – Error logs and Tracefiles – Recovery and Escalation procedures • Connectivity Overview – NDBAPI – Cluster/J – LDAP • Severalnines Tools – Monitoring and Management – Benchmarking – Sandboxes – Configuration and capacity planning • Conclusion Copyright 2011 Severalnines AB Control your database infrastructure 7
- 8. Agenda: Lab Exercises (only applicable to instructor-led training classes) • Lab Exercises – Installing and Loading data into MySQL Cluster – Starting/stopping nodes, recovery – Query tuning – Backup and Restore – Configuration Upgrade • Would you like to try something particular? – This is possible too, speak with your instructor Copyright 2011 Severalnines AB Control your database infrastructure 8
- 9. Prerequisites • Readers / Participants have understanding of SQL and basic database concepts. • Laptops/PCs for hands-on exercises • Linux: 1GB RAM • Windows: 2GB RAM • Approx. 20GB disk space and Virtualbox installed. • Virtualbox can be downloaded for free at http://www.virtualbox.org/wiki/Downloads • MySQL Cluster version 7.1 or later Copyright 2011 Severalnines AB Control your database infrastructure 9
- 10. 5th Installment Severalnines Cluster Self-Training Part 3: Internals Copyright 2011 Severalnines AB Control your database infrastructure 10
- 11. Topics covered in Installment 5 • NDB Design Internals Copyright 2011 Severalnines AB Control your database infrastructure 11
- 12. Section 5: Internals Copyright 2011 Severalnines AB Control your database infrastructure 12
- 13. Software Model • Software model is inherited from Ericsson AXE swtiches. • The kernel is composed of several software blocks • A block owns a functionality – Transaction handling, index handling etc. • Blocks communicates with each other using signals – Signals can also be sent to blocks in other nodes (distributed) • No data sharing between blocks! Copyright 2011 Severalnines AB Control your database infrastructure 13
- 14. Blocks • Blocks are software modules with pre-allocated data structures (dynamic memory allocation will be added later) • No pointers • Logical indexes into data structures with run-time checks of index out of bounds • Each block is a separate C++ class with one generic parent (SimulatedBlock) • Specific entry methods for all incoming signals • A Virtual Machine schedules the execution of signals inside a block. Copyright 2011 Severalnines AB Control your database infrastructure 14
- 15. Blocks Overview • DBACC – Access control, lock manager • DBDICT – Meta data management, table and index defintions • DBDIH – Data distribution management, data fragmentation and replicas, Local Checkpoints (LCP) and Global Checkpoints (GCP), system restart • DBLQH – Local query handler, local transaction management, local data operations • DBTC – Transcation coordinator, distributed transaction management, global data operations • DBTUP – Tuple manager, manages physical storage of data (read, insert, update, delete, and monitoring changes of tuples) Copyright 2011 Severalnines AB Control your database infrastructure 15
- 16. Blocks Overview • BACKUP – On-line backup – LCP • CMVMI – Configuration management, interacting with management server, interaction between blocks and virtual machine • DBTUX – Local management of ordered indexes • DBUTIL – Internal interface to transactions and data operations • NDBCNTR – Ndb Cluster manager, adaption to logical cluster (QMGR), initialization and configuration of blocks Copyright 2011 Severalnines AB Control your database infrastructure 16
- 17. Blocks Overview • NDBCNTR – Ndb Cluster manager, adaption to logical cluster (QMGR), initialization and configuration of blocks • NDBFS – Abstraction layer on-top of local file system • QMGR – Cluster manager, handles logical cluster, cluster membership • RESTORE – Supports restoring on-line backup • SUMA – Subscription manager, data and meta-data event monitoring • TRIX – On-line unique index build Copyright 2011 Severalnines AB Control your database infrastructure 17
- 18. Virtual Machine • The Virtual Machine – Job schedulation mechanism (job== process signal) – Jobs represented by signal buffer (Job Buffer) – Hides OS details from blocks – Single threaded implementation with job switching as concurrency model • very inexpensive context switching • requires that blocks cooperate and relinquish control within a resonable time slot -> more effort for programmer. Copyright 2011 Severalnines AB Control your database infrastructure 18
- 19. Virtual Machine and Signals • Signal types – Local signal • signal to another local block, to be executed when associated job is scheduled by the Virtual Machine Equivilent to function call. No scheduling – Remote (distributed) signal • signal to a block on another node, to be executed when remote associated job is scheduled by the remote Virtual Machine – Delayed signal • to be executed after a certain time, e.g. 10ms. • Signals can be fragmented and be up to 12 GB (non fragmented signals up to 100B) Copyright 2011 Severalnines AB Control your database infrastructure 19
- 20. Sample Signal void Backup::execBACKUP_REQ(Signal* signal) { jamEntry(); BackupReq * req = (BackupReq*)signal->getDataPtr(); const Uint32 senderData = req->senderData; const BlockReferencesenderRef = signal->senderBlockRef(); const Uint32 dataLen32 = req->backupDataLen; // In 32 bit words const Uint32 flags = signal->getLength() > 2 ? req->flags : 2; if( dataLen32==0) {jam(); abort()} if(getOwnNodeId() != getMasterNodeId()) { jam(); sendBackupRef(senderRef, flags, signal, senderData, BackupRef::IAmNotMaster); return; }//if Copyright 2011 Severalnines AB Control your database infrastructure 20
- 21. Signal Types • There are a number of different signal types. Here are som common. – REQ request signals – CONF confirmation signals (ack) – REF refusal signals – REP report signals – ORD order signals (no reply) Copyright 2011 Severalnines AB Control your database infrastructure 21
- 22. Virtual Machine and Signals • select() on all communication interfaces (called transporters in NDB) • Receive Signals from all communication interfaces • Check Timed Signals • Execute Signals • Send Signals in buffers beloning to communication interfaces (transporters). Copyright 2011 Severalnines AB Control your database infrastructure 22
- 23. Scheduler • ThreadConfig.cpp • FastScheduler.cpp • Two levels of scheduling – Scheduling between performing jobs and send/receive on transporters (ThreadConfig.cpp) – Scheduling between jobs of different priorites (FastScheduler.cpp) Copyright 2011 Severalnines AB Control your database infrastructure 23
- 24. Scheduler • Signals have priorities: – Priority A Signals executed first – Priority B Signals executed then – Priority C Signals executed then (not really used) – Priority D, used to buffer Delayed Signals that are then put into B job buffer Copyright 2011 Severalnines AB Control your database infrastructure 24
- 25. Transporters • Transporters facilitates the communication between nodes. – Point to point – each node has one transporter to any other node. – Hides the underlying communication media – Different transporters can have different characteristics (latency, bandwidth, etc.). – Currently TCP/IP sockets, Shared Memory, and SCI Copyright 2011 Severalnines AB Control your database infrastructure 25
- 26. Threads in NDB • Data node consists of a number of threads – Main thread handling the execution (Execution Thread) • Transactions and operations are executed in a single thread • Please note that the Execution Thread is really a set of threads in MySQL Cluster 7.x (when using the multi- thredaded daemon, see next few slides). – Watchdog thread • Makes sure the main thread is not stuck somewhere – Filesystem threads • Handles async i/o such as writing Local and Global checkpoints Copyright 2011 Severalnines AB Control your database infrastructure 26
- 27. Real-time Extensions • The Threads can be bound to CPU cores – reduce context switching • Bind Maintenance threads to one Core – Filesystem threads, watchdog threads – LockMaintThreadsToCPU=<cpuid> • Bind Execution (main thread) to another core – LockExecutionThreadToCPU=<cpuid> • Use cat /proc/interrupts to find out which CPU to avoid – On some Oss CPU 0 is used for interrupt handling of eth0 Copyright 2011 Severalnines AB Control your database infrastructure 27
- 28. Real-time Extensions • SchedulerSpinTimer=200 (us) – Perform select(t=0) (non-blocking) for 200us – Only in MySQL Cluster 6.3 (or non-multithreaded data node) • SchedulerExecutionTimer=50 (us) – Receive and execute more signals before sending – Only in MySQL Cluster 6.3 (or non-multithreaded data node) • RealtimeScheduler=1 Copyright 2011 Severalnines AB Control your database infrastructure 28
- 29. Protocols • MySQL Cluster is a small and fast democracy with a president (master) • Protocols in MySQL Cluster is based on Consensus – President initiates a request and sends to all Participants • (REQ signal) – President expects to get a CONF back from the Participants – If a participant sends a REF, then it will most likely be exclued Copyright 2011 Severalnines AB Control your database infrastructure 29
- 30. Protocols • Central protocols – Two phase commit protocol (2PC) – Global checkpoint protocol (GCP) – Local checkpoint protocol (LCP) – Heartbeat protocol (HB) Copyright 2011 Severalnines AB Control your database infrastructure 30
- 31. Multi-threaded Data Node • From MySQL Cluster 7.0 the data node is multi- threaded – MaxNoOfExecutionThreads • Set to 8 (max) for 8 core machines. – The data node will then have • 1 TC thread • 4 LQH threads (workers) • 1 CMVMI thread (communication) Copyright 2011 Severalnines AB Control your database infrastructure 31
- 32. Multi-threaded Data Node TC TC thread LQH LQH LQH LQH Worker ACC TUP ACC TUP ACC TUP ACC TUP threads P0 P0 P1 P1 P1 P2 P3 P3 Index Data Index Data Index Data Index Data Memory Memory Memory Memory Memory Memory Memory Memory P0 P1 P2 P3 REDO LOG D8 D9 D10 D11 Copyright 2011 Severalnines AB Control your database infrastructure 32
- 33. Multi-threaded Data node • Each worker has one or more partition – Depends on the number of workers: • 1 worker -> 4 partitions / worker • 2 workers -> 2 partitions / worker • 4 workers -> 1 partition / worker • Each partition maps to one Redo log segment • Communication between threads is efficient – Uses the instruction set available in modern CPUs to have “lock free” communication • Each thread has its own scheduler • Typically it is either the TC thread or the CMVMI thread that gets overloaded, unless you use a lot of scans/range scans then it is the Workers that become the bottleneck first. Copyright 2011 Severalnines AB Control your database infrastructure 33
- 34. Coming next in Section 6: Part 4: Installing MySQL Cluster Copyright 2011 Severalnines AB Control your database infrastructure 34
- 35. We hope these training slides are useful to you! Please visit our website to view the next section of this training. For any questions, comments, feedback or to book a training class, please contact us at: services@severalnines.com Thank you! Copyright 2011 Severalnines AB Control your database infrastructure 35
- 36. Disclaimer © Copyright 2011 Severalnines AB. All rights reserved. Severalnines& the Severalnineslogo(s) are trademarks of Severalnines AB. MySQL is a registered trademark of Oracle and/or its affiliates. Other names may be trademarks of their respective owners. Copyright 2011 Severalnines AB Control your database infrastructure 36
Editor's Notes
- Dear Jury, Just as the electric grid revolutionized access to electricity 100 years ago, we at Severalnines believe that Cloud Computing will revolutionize IT where organizations will be able to plug into extremely powerful computing resources over the network. We have already seen the beginnings of this new wave, where the current infrastructure stack is being challenged and disrupted by a whole set of new technologies. For instance, in the database market, over 40 startups have received funding over the past 18 months. Severalnines is not building yet another database product, we believe there are already a lot of good technologies available. To manage a database costs 4 times the purchase price, and yet, very few companies are addressing this problem. Severalnines focuses on solutions to address this underserved segment. The founders of the company have a solid background in databases, having been at MySQL since 2003. The company develops a management platform which is database and cloud agnostic.We are database independent since we do not know who, if anybody, will be the next MySQL of the cloud.We are cloud independent, since we do not want to depend on any cloud vendor (e.g. Amazon or Rackspace) to avoid vendor lock-in. After the Amazon EC2 downtime during the Easter break, hundreds of affected companies have realized the importance of this. There is also a commercial aspect for avoiding vendor lock-in. Severalnines enhances productivity of organizations by attacking the biggest cost associated with database systems. We are a Swedish startup, hosted by SICS in Kista. There is also a small but very efficient development capacity in Singapore. Just as MySQL became a major brand and placed Sweden on the global software infrastructure map, we believe Severalnines can become a serious global player in the emerging Cloud space. We are very thankful that an organization like Eurocloud exists, and would like to thank the jury for considering our application. Kind regards,Vinay Joosery Severalnines AB
- Severalnines has been offering its products free of charge since 2007, while the founders were employed at MySQL. These products are the de-facto standard tools to assist MySQL customers and users in deploying their MySQL clusters. More information about Severalnines at www.severalnines.com
- Severalnines has been offering its products free of charge since 2007, while the founders were employed at MySQL. These products are the de-facto standard tools to assist MySQL customers and users in deploying their MySQL clusters. More information about Severalnines at www.severalnines.com
- Severalnines has been offering its products free of charge since 2007, while the founders were employed at MySQL. These products are the de-facto standard tools to assist MySQL customers and users in deploying their MySQL clusters. More information about Severalnines at www.severalnines.com
- Severalnines has been offering its products free of charge since 2007, while the founders were employed at MySQL. These products are the de-facto standard tools to assist MySQL customers and users in deploying their MySQL clusters. More information about Severalnines at www.severalnines.com