Tag Info

Hot answers tagged

18

OK, let's imagine that you have a distributed database. Let's say you have a node in Oregon and one in California. The CAP theory says that you will run into problems when setting up this type of database. For example, if you query data from one database, it needs to be the same as the data in the other database. This insures that whatever value you have ...


10

BigTable doesn't use SQL (a query language) so SQL can't be used directly to query the database. And BigTable doesn't have "relations" in the same way as relational databases, it's more like bare tables. If you want to get data from two tables, you have to do two lookups, and combine the result set in the application code. In other words the "join" ...


4

You need to talk to your network admin to see if Kerberos delegation is enabled, and whether it's constrained or not (if it's enabled and unconstrained that is a massive potential security risk, just by the by). If it is constrained then you need to register a SPN (Service Principal Name) for your SQL Server and bind your SQL Server account service to that ...


4

If I understand you right, you are going from DB2 10.1 Express-C edition to DB2 10.1 Work Group Server Edition or Enterprise Edition. If that is case then you shouldn't have any issues. The binaries for DB2 are basically the same across editions. It is just a matter of a license file that enables or disables certain features and enables/disables use of ...


4

I have been in a similar situation and highly recommend looking into MongoDB or HBase for your needs. The others (Cassandra/Redis) are good too, but HBase and MongoDB have been battle-tested in large environments, have a large community, good documentation (MongoDB's is better) and generally fulfill the three requirements you are looking for.


3

If you are looking for data models to start with, I'd recommend you take a look at: http://www.databaseanswers.org/data_models/index.htm This is the most comprehensive list of data model examples that I found so far. In fact I have used some samples as starting point for some of my projects. These models won't show you how to take horizontal fragmentation ...


3

How big are the keys? How big are the values? Will the data arrive "continually", or is this a one-time load? Let's see all the representative queries. You have mentioned only SELECT val FROM tbl WHERE key = 'constant'; Are there others ? In general key-value stores suck big time. But maybe we can make your special case work not-too-badly.


3

They are not really the same, because of the scope of the data ACID Atomicity Consistency : All Applied Data Changes Provide Consistent View of Data For All DB Connections Isolation Durability CAP Consistency (All Nodes Have Same Data via Eventual Consistency) Availability Partition-Tolerance : system continues to operate despite arbitrary message ...


2

Postgres-XC does not run on Windows: Postgres-XC currently runs on the Linux Operating Systems running on 64-bit Intel(R) processors. The development team tested this with Cent-OS 5.3 and 5.4. You might be able to build the data nodes and possibly the coordinators on Windows, but the GTM will take major work to port to Windows.


2

HP Vertica has their community edition for download at https://my.vertica.com/download-community-edition/. If you don't have one already, you will be prompted to create a MyVertica login. If you have problems creating a login, please email me at dsandahl@vertica.com. If you have further issues loading your data, you can tweet at @VerticaHelp and support will ...


2

Compare a fragmented distributed database with a replicated distributed database. A replicated database would contain all columns and all records on more than one database instance; whereas, the separate instances of a fragmented distributed database do not all contain the same columns and/or records. Vertical Fragmentation: A vertically-fragmented ...


2

Take a look at the views of Micheal Stonebraker here - pages 21 and 22. Stonebraker is both an academic and a serious commercial player in the database world (Postgres, Vertica, Informix, VoltDB inter alia). For your particular question, note his views on banking. Stonebraker is and advocate of "NewSQL". NewSQL is a two-pronged approach. For OLTP (banking, ...


1

"Is MySQL suitable for distributing or cluster databases ?" While you certainly can do manually sharding, the truth is that the most common MySQL cluster technologies are focused in high availability. NDB Cluster is based mainly on a memory database with lots of throughput, but not necessarily low-latency and large datasets. Galera does not support ...


1

The short answer is: NO, Oracle does not do this in the way you propose BUT: there is a way to emulate this but don't forget that Oracle has a share everything architecture, meaning that all data is available for all instances on all nodes. Using services and application code that connects to specified services running on their own instances can use ...


1

Bill Karwin has already given the best answer that I can think of... but I think some additional explanation might be in order as to why the link you posted in the original question probably does not provide a sufficient solution. Your DBA is, at least in one sense, correct. The problem is that the illustration is deceptive. The three distributed nodes in ...


1

Check out this blog: http://www.mysqlperformanceblog.com/2012/01/11/making-the-impossible-3-nodes-intercontinental-replication/ We will create 3 nodes Cluster with nodes on different continents (Europe, USA, Japan) and each node will accept write queries. The solution used is Percona XtraDB Cluster.


1

The alternative to using database links would be to create different connections in your SSIS package for each database and to use those connections to pull data from the remote database to the SSIS server and then push that data into the target database. It is highly unlikely, however, that this would lead to performance improvements. Most likely, it ...


1

A materialized views is not a replacement for a database link. It would likely create a local materialized view using data pulled accross a database link. What you need to do is tune the query that you are using for your materialized view. Run the query in sqlplus with timing and autotrace turned on. Look for cartesean joins, full table scans and sorts. ...


1

To my knowledge it does not allow that level of parallelism. If that's what you are trying to do the best approach is to go with Postgres-XC or GridSQL, but Postgres-XC while more complex is also more flexible. http://postgres-xc.sourceforge.net/ is the project page.


1

Don't do this with only one slave. If you do you are making your application more subject to failure, not less. Basically if you do this, and you lose your slave, your master will hang. You say your backup box is crappy, and I assume that means reliability-wise too, so this is more likely than losing your master. This is probably not what you want. Do ...



Only top voted, non community-wiki answers of a minimum length are eligible