硬件压缩在网络应用程序中扮演什么角色
很长一段时间以来，我一直认为当我通过网络发送数据时，数据会被压缩，因为网卡几十年来一直支持硬件压缩。然而，最近的测试表明这是一个幼稚的观点，因此我开始在我的应用程序和服务中实施软件压缩。
因此，我的问题是，作为软件开发人员，我们可以在多大程度上依赖硬件网络压缩，或者我是否从根本上误解了它的用途。
For a long time I've assumed that when I send data down the wire, it'll be compressed for me, as NICs have supported hardware compression for decades. Recent testing however has shown this to be a naive view, so have started implementing software compression in my apps and services.
My question therefor is, to what extent can we as software dev's rely on hardware network compression, or have I fundamentally misunderstood what it's there for.
如果你对这篇内容有疑问，欢迎到本站社区发帖提问 参与讨论，获取更多帮助，或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱，如果其他用户或者作者回复了您的评论，将不能在第一时间通知您！
发布评论
评论（1）
有 TCP/IP 标头压缩（不是有效负载）。思科还提供第 2 层有效负载压缩 http://www.cisco。 com/en/US/tech/tk713/tk802/technologies_q_and_a_item09186a008019be75.shtml 在网络堆栈中非常低。
对于网络应用程序，您位于第 7 层，这些压缩对您来说应该是透明的。根据有效负载的性质，您可能会获得比第 2 层（例如 H.264）更高的吞吐量，或者略有改进（例如文本文件传输）。由于下层是透明的，我建议不要赌它。您可能正在通过根本不压缩的东西进行传输（例如，ATM 上的 IP），因此您无需依赖“硬件压缩”。
There are TCP/IP header compressions (not the payload). Cisco also has Layer 2 payload compressions http://www.cisco.com/en/US/tech/tk713/tk802/technologies_q_and_a_item09186a008019be75.shtml that's very low in the network stack.
For network apps, you're at layer 7, and these compressions should be transparent to you. Depending on the nature of your payload, you may have a much superior throughput than what's done in layer 2 (e.g. H.264), or marginal improvement (e.g. text file transmission). Since lower layer is transparent, I'd suggest don't bet on it. You might be transmitting over something that does not compress at all (e.g. IP over ATM) and thus there's nothing for you to rely on "hardware compression".