Hi Ilya, thanks for your feedback. Em seg., 24 de abr. de 2023 às 15:28, Ilya Maximets <[email protected]> escreveu:
> On 4/13/23 13:35, Roberto Bartzen Acosta wrote: > > Hey folks, > > > > I've been reading some docs about parser acceleration for json and I > thought of the ovsdb case. I already had serious performance issues with > ovsdb in large-scale deployments. I noticed was that most of the processing > was related to ovsdb tasks. > > > > Do you have any idea if any acceleration routine using vectorized > classification concepts/SIMD as presented in this article [1]/project[2] > would be applicable to the case of JSON RPC parsers? > > > > Maybe Ilya, who knows more deeply about ovsdb approach and who evaluates > performance using the ovn-heater can contribute here. > > > > [1] - https://arxiv.org/abs/1902.08318 <https://arxiv.org/abs/1902.08318 > > > > [2] - https://github.com/simdjson/simdjson < > https://github.com/simdjson/simdjson> > > > > Kind regards, > > Roberto > > Hi. Thanks for interesting links! > > It might certainly be possible to improve the performance of a JSON > parser with more compiler hacks and SIMD instructions. However, in > my experience, the parser itself doesn't really show up in testing > as a big performance bottleneck. Especially after some good changes > made last year by Rosemarie: > > https://github.com/openvswitch/ovs/commit/2f16123c1bf0793381683ede3d63ae6da379d9b7 > So, making it faster may not worth the cost of increased code complexity. > The main challenges in ovsdb-server these days are mostly related to > filtering large amounts of data and sending it out, not really the > JSON parsing, AFAIK. > Cool! It's a great improvement included in version 3 (I haven't had an opportunity to test this version on a large scale setup). I'm also very interested in the latest improvements you and Han are working on - especially with Han using ChatGPT to discuss computational complexity questions ;) > > Best regards, Ilya Maximets. > Best regards, Roberto -- _‘Esta mensagem é direcionada apenas para os endereços constantes no cabeçalho inicial. Se você não está listado nos endereços constantes no cabeçalho, pedimos-lhe que desconsidere completamente o conteúdo dessa mensagem e cuja cópia, encaminhamento e/ou execução das ações citadas estão imediatamente anuladas e proibidas’._ * **‘Apesar do Magazine Luiza tomar todas as precauções razoáveis para assegurar que nenhum vírus esteja presente nesse e-mail, a empresa não poderá aceitar a responsabilidade por quaisquer perdas ou danos causados por esse e-mail ou por seus anexos’.*
_______________________________________________ discuss mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-discuss
