IBM Informix Dynamic Server (IDS) needs careful handling of data manipulation language (DML) operations affecting millions of rows. If performed in a single SQL statement on a logged database, the database engine must keep each affected row locked until it completes: even if no transaction has been started with BEGIN WORK, the statement still runs inside an implicit transaction. Assuming the table has LOCK MODE set to ROW (the norm for OLTP) rather than PAGE (usually avoided due to concurrency problems), the result is millions of locks, particularly as they are required on each index as well as the table. The documentation for the LOCKS configuration parameter, which defines the initial size of the lock structure in the resident shared memory segment, states that each requires 100-200 bytes depending on the platform. Should it run out and need to allocate an extension lock structure in virtual shared memory, it may have to add new segments dynamically totalling gigabytes. Unless the Enterprise Edition is in use, this may cause user sessions to be starved of memory or even rejected. Following a feature request by the author, configuration parameter SESSION_LIMIT_LOCKS is available from IDS 12.10.xC4 that you can use to prevent an excessive number of locks by any one session, but that will obviously result in the statement being aborted if the limit is reached. Locks can be avoided entirely by placing an EXCLUSIVE lock on the table or by temporarily disabling logging on the database or altering the type to RAW, but this will most likely interfere with other users, break applications, and would invalidate any replication. There is a second major problem with very large transactions: if so much data is affected that most of the logical log space has been consumed, depending on configuration parameter LTXHWM, the statement will encounter “Long Transaction Aborted”. It can then take at least as long for the transaction to be rolled back, during which time the instance may be unusable if LTXEHWM has also been reached. The right way to avoid this entirely is to split DML statements into smaller transaction affecting only a few thousand rows each at most. The two most common scenarios are when loading fresh data from a file or deleting a large number of rows. These can be safely achieved committing a few thousand rows per transaction with the dbload tool provided with IDS (see wrapper script at the end of this article) and the dbdelete open source ESQL-C program by Art Kagel (see also a Python version by Andrew Ford). Other scenarios may need to be specifically coded. Informix stored procedure for mass delete allows millions of rows to be safely deleted in one operation inside any SQL session. This article describes a stored procedure alternative to dbdelete based on it.

[Read More…]

With effective RAM disk and Informix management, you can make your disk space work harder. Most editions of Informix Dynamic Server limit the amount of shared memory that can be allocated: you can compare the Informix Version 12 editions here. This primarily constrains how big the buffer pools (disk cache) can be, while still leaving enough for other essential memory pools. However, modern machines will often have much more RAM than this, which could be put to better use. This article will provide the complete process to use spare memory as RAM disk file systems so you can get more out of your RAM disk and Informix. Their contents are volatile, so they can only be used for temporary tables (DBSPACETEMP) and sort/merge files (PSORT_DBTEMP). The commands to create them on all supported Linux and UNIX flavours will be given, along with other relevant environment and configuration parameter settings. Temporary data is then never written to disk, dramatically improving run times of larger queries and preventing interference with OLTP sessions by reducing buffer turnover. In an actual case, disk writes were reduced by 96%, eliminating SAN contention with other applications. Even using Ultimate Edition which has no shared memory limit, index builds in temp dbspaces are not only slower but can crash the instance on some versions if they run out of space. This can be avoided by setting PSORT_DBTEMP to use file systems instead, and run time can be made shorter still if these are RAM disks. Note that RAM disk should not be confused with solid state drives (SSD), which are persistent and a better technology for logical and physical logs when combined with RAID 1 or 10.

[Read More…]

Analysis of SQL statements going through a database engine can be the most important task to improve user response times. Even if you think all is well, you may discover coding faults or unexpected choices made by the query optimizer, resulting in longer execution times and higher system load, that can often be easily fixed once you know which are the worst. It isn’t just the longest queries that matter: saving a few milliseconds on a statement can have a big impact if it’s run thousands of times. Increasingly, database security and auditing is becoming a focus. For example, there may be a requirement to identify which users have made schema changes or updated certain tables. Most auditing solutions focus on how specific data records have changed, however, sometimes it is more meaningful to know what SQL was actually run to generate that change. Whatever the requirement, in order to analyse SQL workload, we need to capture SQL statements. Over recent years, a number of commercial solutions have entered the market that are designed to provide a SQL capture capability. We’ve evaluated most of these products including: iWatch (Exact Solutions); SQL Power (SQL Power Tools). An analysis of these tools is outside the scope of this particular article. This article will focus on what can be achieved with the underlying Informix software utilities (including Informix SQL/SPL scripting approaches) and various Informix management interfaces and tools.

[Read More…]

Fragmentation vs Sharding. So, what is it? Fragmentation has been available in Informix since V7.00 which was released in the mid-nineties. It allows you to group data rows and/or indexes for a table according to a user-defined distribution scheme and physically place these fragments in separate dbspaces and on separate physical disks. This can bring benefits to query performance by spreading the I/O over multiple devices, by elimating the need to read fragments not relevant to a particular query or even scanning multiple fragments in parallel. As data volumes grow, the ability to fragment large tables across multiple dbspaces can also reduce the requirement to create dbspaces with larger page sizes and the additional buffer pools required for them. But, in today’s Big Data era, as data storage requirements grow at an ever increasing pace, what if the performance and capacity of a single server can no longer meet these demands ? One possible answer could be Sharding. Sharding was introduced at V12, it allows you to group data rows and index keys for a table according to a user-defined distribution scheme and physically place these fragments on separate servers, locally or remotely. This allows the resources of some or all of these servers to be used when processing queries. As your database grows, rather than scaling up by adding more processors and/or RAM to an existing server, you can scale out by adding more servers. Also, as Sharding makes use of Informix Enterprise Replication, there is no requirement for the server hardware and operating systems to be the same. Read on to discover the pros and cons of Informix V12 Fragmentation vs Sharding.

[Read More…]

This article follows on from the earlier TPC-C benchmarking performed on IDS 12.10, if you missed it, you can read it here. This article takes a view on the topic of DIRECT_IO and its use within IDS, it also highlights the impact of file system caching on Linux, which is relevant for those editions of Informix where DIRECT_IO cannot be used. Lastly, a ‘just for fun’ comparison of DIRECT_IO and RAW devices for readers to come to their own conclusions. Read on to learn more about the impact of DIRECT_IO and File System Caching on IDS.

[Read More…]