Re: Informix evaluation

In article <32B9D4FD.375E@alliedsignal.com>, Naushi Hussain <hussain@alliedsignal.com> writes >We are doing an eval. of some high end DBMSs for a MPPS applications. If someone >could help me define these terms ( a line each) and why would they be necessary >to consider in evaluating a DBMS. > >' Async I/O > The database server does not have to wait for a disk read/write to complete but can do something else whilst the disk read/write is happenning. Provides up to 40% better performance. >' Virtual Processor Part of multi-threading which means less UNIX processes are required. This means less memory is required. Also less swtiching between processes means better performance. >' Dynamic Shared Memory Shared memory means different UNIX processes have access to the same area of memory. Provides fast communication between processes. Dynamic means memory usage increases as required - no fixed limit. > >' 64 Bit support > A single process can access 2 to be power 64 bytes of memory. Means you can have more than 4 gigabytes of memory on the machine. > >' Share nothing implementation Data is not held centerally and passed from machine to machine across the memory. i.e. machine do not share data. Reduces network usage a lot and provides much better performance. >' Function Ship VS Data ship Same thing expalined in a different way. Data is not shipped from one machine to another rather the machine which has the data on it is told to processes the data. >' Parallel execution of functions like DML, utilities Parallel processing. A task is split into smaller subtasks which can be run on several CPUs/machines at the same time. Provides much better performance. > >' Intelligent Data Partitioning > Data can be distributed across machines in a way that reflects how it is used i.e split the data across several machines/disk drives and have each machine proccess it's own part at the same time. Increases performance. > >' log based application Transaction logging is used to allow recovery is case of system failure and make sure an update is not partly written to the database. >' sync - 2 phase commit 2 phase commit helps to ensure that updates complete correctly when several machines are involved. i.e. handle failure of one machine gracefully. > > >' centralized console Allow adminstration of several machines from one terminal. > >' Security > ' Roles Rather than define security access for each user define security access for a group of users (a role) and then assign the user that role. Simplifies administration when you have lots of users. You can group the users together and define security access for the group. > ' Audits Allow auditing i.e when a change is made to the database keep a record of who changed it and what they changed. > ' Dynamic allocation Security can be configured by the database administrator and is not fixed when an application is written. i.e. permission to look at or modify something in the database can be changed without have to get a programmer to change the application. > >' Multiple datatype support > ' blobs Binary Large Objects, can be used to store any type of data. Not just numbers and letters but things like sound and video. > > >' Year 2000 compliance > Means dates are stored as 4 characters - 1901 rather than just 2 characters (01). Databases which are not Year 2000 compliant assume the century is 19 and only store 2 characters i.e. 01. These would then not work in the year 2000 and beyond. > >' Robust procedural language > ' Triggers & stored procedures Allow actions (procedures) to be defined within the database itself. These can either be used by applications or within triggers. Trigger are ways to define what actions (procedures) are taken when something within the database changes. > ' extended SQL Database modifications and searching are done using a language called SQL (Structured Query Language). There is a standard for this language known as the ANSI standaed. Extended means additions to the language which are not defined in the ANSI standard. These usually either simplify application design or increase application performance. >' Performance Management > ' advanced indexing Indexes improve application performance in the same way that an index in a book allows you to find a particulat piece of information quickly. Advanced indexing means using the latest methods of indexing to improve performance. > ' cost based optimizer Queries (seaches) can often be performance in several different ways. An optimizer find the optimal (fastest) method. Cost bases means it uses a method to predicate the cost (speed) of each method bu storing information about the data which is being stored E.g. how much data, how many duplicate values are in the data. Rather than a rule based optimizer which uses fixed rules about which is the best method and does not look at the data which is being stored. > -- David Williams