Still time to attend The Next Database Platform 2020 online for free: Sign up and dive into the next era of info storage
(2020/08/26)
- Reference: 1598478809
- News link: https://www.theregister.co.uk/2020/08/26/the_next_database_platform/
- Source link:
Event The IT world would have been a far simpler and easier place if the relational databases commercialized in the 1980s and expanded in the 1990s and 2000s had absorbed new data types quickly and efficiently while at the same time scaling up in larger machines and scaling out across multiple machines.
To their credit, the remaining major relational databases used by enterprises – Oracle, IBM DB2, Microsoft SQL, and MySQL and PostgreSQL in their various guises – have done a pretty good job of absorbing object and XML and JSON document formats as well as adding columnar data store and in-memory options.
But invariably, these relational databases come to a breaking point where they can’t get answers fast enough, they can’t scale across enough compute and storage to hold extremely large database, or both. And they are always – always – very expensive. The cost of the database software can rival that of the compute, storage, and networking that underpins the database, and then there are always supplemental costs for add-ons, such as caching and messaging interfaces that attempt to speed things up.
And so, as the types of applications and the types of data that are being stored is increasing, and demands on lowering latency and increasing scale are relentless, it is no surprise then that there is a true Cambrian explosion in the database industry over the past several years. We not only have NoSQL and NewSQL databases that emerged a decade ago because of the limits of legacy relational databases, but we have a whole new crop of databases that store information in time series, graph, object, document, and relational formats, with varying degrees of structure and schema.
This is equivalent to the x86 processor taking on mainframes, proprietary minicomputers, and RISC/Unix servers in the early 1990s, which obviously had a huge impact on the modern data center.
Because performance matters, many of these databases can run in-memory and a number of them can be accelerated by flash or 3D XPoint storage or by adjunct compute engines such as GPUs or FPGAs. And more than a few of them have automagic scaling and data partitioning (important for data sovereignty reasons) across vast geographical instances. And, here is the important part, many of these new database alternatives are considerably cheaper to acquire than those legacy relational databases. This is equivalent to the x86 processor taking on mainframes, proprietary minicomputers, and RISC/Unix servers in the early 1990s, which obviously had a huge impact on the modern data center.
But rather than creating a single substrate of general purpose compute that drove up volumes and drove down prices as the x86 engine did in the data center, this database revolution is spawning variety in its splendor, and spurring competition that is driving down prices. It doesn’t hurt that to get early adopters, these companies have to offer very attractive pricing for a given amount of structured or semi-structured data to be stored compared to those legacy databases because the risk of changing databases is so large that the reward has to be great.
For those who need lower latency or higher scale than a legacy relational database can provide, they can be charged a premium – and will pay it, too, because they need to solve their latency and scale problems.
There is a new era in databases, and we are thrilled to be exploring it with you.
So join us for The Next Database Platform 2020 on Thursday, August 27 at 12pm Eastern Time, [1]which you can register for – for free – right here .
[1] https://www.nextplatform.com/2020/07/20/the-next-database-platform/
To their credit, the remaining major relational databases used by enterprises – Oracle, IBM DB2, Microsoft SQL, and MySQL and PostgreSQL in their various guises – have done a pretty good job of absorbing object and XML and JSON document formats as well as adding columnar data store and in-memory options.
But invariably, these relational databases come to a breaking point where they can’t get answers fast enough, they can’t scale across enough compute and storage to hold extremely large database, or both. And they are always – always – very expensive. The cost of the database software can rival that of the compute, storage, and networking that underpins the database, and then there are always supplemental costs for add-ons, such as caching and messaging interfaces that attempt to speed things up.
And so, as the types of applications and the types of data that are being stored is increasing, and demands on lowering latency and increasing scale are relentless, it is no surprise then that there is a true Cambrian explosion in the database industry over the past several years. We not only have NoSQL and NewSQL databases that emerged a decade ago because of the limits of legacy relational databases, but we have a whole new crop of databases that store information in time series, graph, object, document, and relational formats, with varying degrees of structure and schema.
This is equivalent to the x86 processor taking on mainframes, proprietary minicomputers, and RISC/Unix servers in the early 1990s, which obviously had a huge impact on the modern data center.
Because performance matters, many of these databases can run in-memory and a number of them can be accelerated by flash or 3D XPoint storage or by adjunct compute engines such as GPUs or FPGAs. And more than a few of them have automagic scaling and data partitioning (important for data sovereignty reasons) across vast geographical instances. And, here is the important part, many of these new database alternatives are considerably cheaper to acquire than those legacy relational databases. This is equivalent to the x86 processor taking on mainframes, proprietary minicomputers, and RISC/Unix servers in the early 1990s, which obviously had a huge impact on the modern data center.
But rather than creating a single substrate of general purpose compute that drove up volumes and drove down prices as the x86 engine did in the data center, this database revolution is spawning variety in its splendor, and spurring competition that is driving down prices. It doesn’t hurt that to get early adopters, these companies have to offer very attractive pricing for a given amount of structured or semi-structured data to be stored compared to those legacy databases because the risk of changing databases is so large that the reward has to be great.
For those who need lower latency or higher scale than a legacy relational database can provide, they can be charged a premium – and will pay it, too, because they need to solve their latency and scale problems.
There is a new era in databases, and we are thrilled to be exploring it with you.
So join us for The Next Database Platform 2020 on Thursday, August 27 at 12pm Eastern Time, [1]which you can register for – for free – right here .
[1] https://www.nextplatform.com/2020/07/20/the-next-database-platform/