RobinHood: tail latency aware caching – dynamic reallocation from cache-rich to cache-poor
RobinHood: tail latency aware caching – dynamic reallocation from cache-rich to cache-poor Berger et al., OSDI’18
It’s time to rethink everything you thought you knew about caching! My mental model goes something like this: we have a set of items that probably follow a power-law of popularity.
We have a certain finite cache capacity, and we use it to cache the most frequently requested items, speeding up request processing.
Now, there’s a long tail of less frequently requested items, and if we request one of these that’s not in the cache the request is going to take longer (higher latency). But it makes no sense whatsoever to try and improve the latency for these requests by ‘shifting our cache to the right.’
Hence the received wisdom that unless the full working set fits entirely in the cache, then a caching layer doesn’t address tail latency.
So far we’ve been talking about one uniform cache. But in a typical web application one incoming request might fan out to many back-end service requests processed in parallel. The OneRF page rendering framework at Microsoft (which serves msn.com, microsoft.com and xbox.com among others) relies on more than 20 backend Continue reading
The Mesosphere Kubernetes Engine allows for multi-cloud and infrastructure management from a single pane of glass and reduces cluster sprawl.
The chip giant’s third-quarter fiscal 2018 revenue of $19.2 billion was an all-time record, up 19 percent year over year, with all business groups posting record revenue.
The London-based telecom provider’s current Chief Executive Gavin Patterson announced in June that he would leave BT amid company struggles.
VMware has more than 2,000 SD-WAN customers, while the second and third ranked SD-WAN vendors count about 1,000 customers, each.
The president issued a memorandum that is intended to make more spectrum available so that operators can more quickly deploy 5G.
CEO Rajeev Suri told investors the company need to shave nearly $800 million in costs by the end of 2020.
The company admits that Kubernetes might not be the answer for managing edge deployments. But it says it's important to get an open source option out now ahead of 5G deployments.
CEO Jeff Ready expect this to be the last round before reaching profitability. But he won’t comment on IPO plans.
IT organizations complain that the lack of SDN-specific skills among engineers is resulting in a slow transition to virtualization but engineers don’t have much incentive to get those skills.
It's been more than six years since NFV became a reality but for service providers the transformation is far from over.