How to provide cloud data at ultra-high speed for advanced analytics

To fuel advanced analytics applications, extraordinary powerful computing resources are needed as well as massive amounts of data served at ultra-high speeds. Modern multi-core CPUs do support the extreme computing power needed. The ability to cost-efficiently provide both access to data at ultra-high speed and computing with extreme I/O power is crucial to secure the benefits and the value of using advanced analytics applications.

Here is the dilemma: storage solutions - and in particular cloud-based storage solutions -  do not provide the access speeds needed to support applications and use cases like advanced analytics, ML and AI.

ZebClient provides a solution to that by serving data at ultra-high speeds to applications from the ZebClient acceleration layer.

To enable advanced analytics, the underlying infrastructure needs to not only deliver super-high data access speeds but also give access to extremely large amounts of data. The workloads of advanced analytics are heavy – and they are not homogeneous. Your infrastructure needs to support extremely high throughput and deliver extremely low latency for mixed workloads where both random and sequential read/write operations occur and data files vary in size from extremely small to super large.

The ZebClient acceleration layer speeds up the access to cloud data to match the requirements of high-performance compute  like AI, ML and advanced analytics. ZebClient keeps data and metadata in the acceleration layer and close to applications as long as data is in use. When data is not actively used by applications anymore, ZebClient evicts data to low-cost cloud storage.

The ZebClient acceleration layer is based on a set of compute clusters that run data management and metadata management workloads. Using either the ZebClient ready-to-go designs or a customised setup, the acceleration layer is configured and dimensioned according to the specific needs of the applications in use. The acceleration layer can be installed either on-premise, in a private cloud or in public cloud.

As a result, ZebClient enables the use of cloud-based data lakes  - even for the most demanding high-performance compute applications.