codecentric mittendrin

Meetups, Stammtische, Hackathons, User Groups: Die codecentric ist weit mehr als die Summe ihrer Mitarbeiter und Projekte.

Hinter jeder erfolgreichen Software steht eine starke Community

Wissensvermittlung, Nachwuchsförderung und der fachliche Austausch auf Augenhöhe sind für uns Herzensangelegenheiten. Wir sind der Meinung: Geteilte Innovationsfreude ist doppelte Innovationsfreude.

Deshalb mischen sich codecentric-Mitarbeiter und -Mitarbeiterinnen gerne unter die Community – ob als Gastgeber, Redner oder Organisatoren diverser Veranstaltungen. Treffen Sie uns auf einem der folgenden Events!

Munich NoSQL

Munich NoSQL

Elsenheimerstraße 55A, München, Deutschland 09.05.2019 | 18:00

Hands on Cassandra and Durable Storage Systems

Alok Dwivedi, Senior Consultant of Instaclustr, will talk about „optimised Java Applications for C*“.

He will walk you through techniques around developing an effective Cassandra client application that is performing and fault-tolerant. We will explore various options available in Cassandra Java Driver that can help achieve these goals.

Cassandra is a highly available, linearly scalable, fault-tolerant database that can offer extremely high throughput and low latency. It has now become de-facto choice for applications that demand high availability, fault-tolerance and performance. But the effectiveness of your Cassandra solution can be limited if the client application doesn’t access Cassandra database in an efficient manner.

The talk will include:

– Creating a robust cluster and Session object taking into account options
like pooling, load balancing, reconnection policy.
– Retry policy and Query idempotence.
– Prepared statements.
– Handling avoidable tombstones.
– Handling data consistency across multiple tables while maintaining

Oleksandr Petrov will talk about „Storage Systems – The Old, The New and The Overlooked“.

The material on storage systems primarily covers two types of storage: B-Trees and LSM Trees. We can contrast these two in terms of buffering and immutability, and it starts looking like we can’t have immutability without buffering and vice versa. Digging into academic and industry works, we quickly find out that we can, and we find that there’s a third vector: data ordering.

It turns out that we can build about a dozen different storage systems mixing and matching buffering, immutability and ordering in different ways: BW-Trees, FD-Trees, Two- and Multi-Component LSM, Bw-Trees and many others.

Starting from the classic B-Trees from 1979 to the era of Log-Structured storage in 199x, and continuing into 2019, we can reproduce the line of thinking that brought the industry where we are today and discover algorithms that give us better solutions to the storage problems we have now.

By revisiting this evolution of thinking and systematizing the findings, we improve our understanding of the modern (and some useful yet overlooked) approaches to durable storage, learn how it works and can think of what lays ahead of us when NVMe and byte-addressable Storage becomes a reality.



Sebastian Frerichs