Unfortunately I didn't find any available seats to take notes for this but this morning a very interesting keynote took place. Representatives from 7 large companies mentioned in the title gathered on stage and answered various questions by MySQL's Kaj Arno.

These questions included things like "how many MySQL servers do you have", "how many DBAs", etc. It was a lot of fun, hopefully someone (Sheeri) will edit and post the video soon.

Keith has a nice summary of everything that went on together with the numbers here.

Update: Venu has even better notes here.

Similar Posts:

• Must-Know People In The MySQL Field …

• Paul McCullagh presents
• BLOB
  • invented by Jim Starkey
  • Basic Large OBject
  • Binary Large OBject
  • photos, films, mp4 files, pdfs, etc
• how MySQL handles BLOBs
  • mysql client send buffer -> receive buffer on the server (max_allowed_packet)
  • streaming a BLOB
    • continuous data stream
    • stream BLOB data directly in and out of the database
    • store BLOBs of any size (>4GB) in the database
    • create a scalable back-end that can handle any throughput and storage requirements. Wouldn't need to know in advance how big the database will get
    • provide an open system that can be used by all engines
    • provide extensions for …

• Jay Pipes, Tobias Asplund
• Finding out the number of rows that would have been returned (MyISAM and InnoDB)
  • SQL_CALC_FOUND_ROWS and FOUND_ROWS()
  • COUNT(*)
  • MEMORY table
  • if query cache is on, then it makes no difference
  • if it's off
    • Memory MyISAM is fastest
    • FOUND_ROWS() is slightly slower than count(*)
  • more in the slides that I'll add later
• quite a lot of humor, these guys are fun
• query union vs index_merge union
  • SELECT … WHERE a UNION SELECT … WHERE b
    vs
    SELECT … WHERE a AND b
  • index_merge wins
• composite index vs index merge
  …

• Phil Hilderbrand of thePlatform for Media, Inc presents
• classic partitioning
  • old school - union in the archive tables
  • auto partitioning and partition pruning
  • great for data warehousing
  • query performance improved
  • maintenance is clearly improved
• design issues in applying partitioning to OLTP (On-Line Transaction Processing)
  • often id driven access vs date driven access
  • 1 big clients could be 80% of the whole database, so there's a difficulty selecting partitioning schemes
• partitioning is only supported starting from MySQL 5.1
• understanding the …

• Robert Hodges from Continuent presents
• About Continuent
  • leading provider of open source database availability and scaling solutions
• solutions
  • uni/cluster - multi-master database clustering that replicates data across multiple databases and load balances reads
  • uses "database virtualization"
• scale-out design motivation
  • protection from db and site failures
  • continuous operation during upgrades
• how come not everyone has it already?
• creating identical replicas across different hosts is hard
  • Brewer's conjecture
• trade-offs
  • DDL support
  • inconsistent reads between replicas
  • deadlocks
  • sequences …

• Suicide
  • having no backups
  • depending on slaves for backup
  • keeping backups on same SAN
  • having a single DBA - Frank didn't like this one at all
  • not keeping binlogs
• Restoring from backup
  • how much time?
  • uncompressed backup ready to mount?
  • separate network for recovery?
• In Fotolog, 1TB of data was severely hit.
  • first problem: backup was highly compressed (tar.gz)
  • uncompressing took hours
  • so keep uncompressed backups (at least last N days)
  • it should be mountable, rather than transferable
• Frank going over recovery modes at …

I had a chance to visit the Kickfire booth after the keynotes and before the first presentation. They gave me a kicking t-shirt, followed by a presentation on the newly announced Kickfire appliance (now in beta, shipping in Fall 2008). Here are some notes I jotted down:

• von Neumann bottleneck
• SQL chip (SQC), packs the power of 10s of conventional CPUs
• Query parallelization on the chip
• On-chip memory - 64GB. No registers - no von Neumann bottleneck
• Beats the performance of a given 3 server, 32 CPU, 130TB box (1TB of actual data - space is used for distributing IO)
• SQC uses column-store, compression, intelligent indexing
• SQL Chip, PCI connection, plugs into a Linux server
  • SQL execution
  • Memory management
  • Loader acceleration
• KDB (Kickfire storage engine), plugs into MySQL …

• Baron Schwartz presents
• only works for SELECTs
• nobody dares admit if they've never seen EXPLAIN
• MySQL actually executes the query
• at each JOIN, instead of executing the query, it fills the EXPLAIN result set
• everything is a JOIN (even SELECT 1)
• Columns in EXPLAIN
  • id: which SELECT the row belongs to
  • select_type
    • simple
    • subquery
    • derived
    • union
    • union result
  • table: the table accessed or its alias
  • type:
    • join
    • range
    • …
  • possible_keys: which indexes looked useful to the …

• Robin Schumacher
• gives overview of MySQL products
• MySQL Enterprise
• MySQL 5.1 announced
  • table/index partitioning -> great for data warehouses, range, cache, key, list, composite, subpartitioning. Partition pruning. Response time greatly improved with proper partitioning.
  • row-based/hybrid replication -> safer and smarter
  • disk-based cluster -> supports bigger DBs
  • built-in job scheduler -> simplified task management
  • problem SQL identification -> easier troubleshooting. Dynamic query tracing is now available, no need to trace things in slow query logs.
  • faster full-text search -> 500% increase in some cases
  • 5.1.24RC available for the conference
• MySQL 6.0
  …

• Speaker: Mikael Ronstrom, PhD, the creator of the Cluster engine
• Explains the cluster structure
• Aspects of performance
  • Response times
  • Throughput
  • Low variation of response times
• Improving performance
  • use low level API (NDB API), expensive, hard
  • use new features in MySQL Cluster Carrier Grade Edition 6.3 (currently 6.3.13), more on this later
  • proper partitioning of tables, minimize communication
  • use of hardware
• NDB API is a C++ record access API
  • supports sending parallel record operations within the same transaction or in different transactions
  • asynchronous and synchronous
  • NDB kernel is programmed entirely asynchronously
  …