OLD VERSION: Scalable and Low Latency Lock-free Data Structures in C++ - Alexander Krizhanovsky

Scalable and low-latency lock-free data structures in C++: Learn how to design and implement efficient, concurrent data structures for high-performance systems, overcoming common challenges and limitations.

Key takeaways

Scalable and low-latency lock-free data structures are crucial for high-performance systems.
Hash tables are more scalable than binary trees, but can lead to collisions.
To mitigate collisions, a bucket tree can be used, where each bucket is a fixed-size memory block.
A balancing trie can be used to balance the tree and reduce cache misses.
The use of metadata can introduce additional memory access and slow down the system.
In log-free design, live-looking can occur, where a record is deleted and then reinserted, causing a collision.
To avoid live-looking, a technique called “rehashing” can be used, where the data structure is reorganized and rehashed.
In a log-free system, it is important to keep the data structure in a consistent state, even in the presence of concurrent updates.
The use of atomic operations can help to ensure that the data structure remains consistent.
In a system with a large number of threads, it is important to use a memory allocator that can handle the memory requirements of the system.
The use of a memory allocator that can handle memory fragmentation can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a scheduling algorithm that can handle the scheduling requirements of the system.
The use of a scheduling algorithm that can handle the scheduling requirements of the system can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a synchronization mechanism that can handle the synchronization requirements of the system.
The use of a synchronization mechanism that can handle the synchronization requirements of the system can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a communication mechanism that can handle the communication requirements of the system.
The use of a communication mechanism that can handle the communication requirements of the system can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a data structure that can handle the data requirements of the system.
The use of a data structure that can handle the data requirements of the system can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a synchronization mechanism that can handle the synchronization requirements of the system.
The use of a synchronization mechanism that can handle the synchronization requirements of the system can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a communication mechanism that can handle the communication requirements of the system.
The use of a communication mechanism that can handle the communication requirements of the system can help to improve the performance of the system.
In a system with a large number of threads, it is important to use a data structure that can handle the data requirements of the system.
The use of a data structure that can handle the data requirements of the system can help to improve the performance of the system.

OLD VERSION: Scalable and Low Latency Lock-free Data Structures in C++ - Alexander Krizhanovsky

More talks