Fastly Now Offering Service and Support Plans
Fastly, the CDN that markets itself as “built by developers, for developers,” has now released several options for service and support plans. These plans aim to help companies ease the transition to Fastly and utilize all of Fastly’s services. Their Professional Services Packages include a plan to assist with the initial configuration and implementation of the Fastly CDN as well as a plan to help a company’s support team maximize the Fastly product and access their team of technical experts.
Fastly’s Support plans are offered at Standard, Gold and Platinum levels. The standard support plan, which is included in signup at any level, provides the basic support that Fastly has previously offered customers. This includes access to the community forum and documentation along with technical support via email. Priority routing and 24/7 access to the support team is available through the Gold Plan, and the Platinum plan allows 24/7 phone and email access to support with a dedicated account manager and availability of a support Slack channel. Fastly’s service and support plans arrive on the market as the popularity of both software-as-a-service and demand for enhanced capabilities and simplicity continue to grow throughout the tech sector.
Google’s DeepMind Switches From Torch 7 to Google’s TensorFlow
DeepMind, founded in 2010, was acquired by Google in 2014. DeepMind has used the machine learning library and open source software, Torch7, for nearly 4 years. DeepMind has accessed the Torch7 open access library to facilitate its learning and since 2014, Google has assisted with bug fixes in Torch7, maintained crucial components, and has been a key contributor.
Google released its second version of its own open source machine learning library, TensorFlow, in November 2015. DeepMind has now made the switch to TensorFlow as its machine learning library. Google expects TensorFlow will enhance DeepMind performance and accelerate its innovation since it operates at a faster pace and on a larger scale than Torch7, which is still used by Facebook, Twitter and many others.
HTTP/2 Server Push Now Available on CloudFlare
CloudFlare launched HTTP/2 support last November and now serves more than 70% of the more than 88k Alexa 2 million websites that are HTTP/2-enabled. CloudFlare has been testing Header Compression since February, but only now is the main acceleration component of HTTP/2, Server Push, available for CloudFlare users. Header compression has been shown to reduce header size an average of 30%, and efficient use of Server Push can result in a 45% performance improvement on a mobile network. Once a link header to the response is formatted, HTTP/2 Server Push can bundle assets when delivering a webpage to speed up performance by eliminating the need for another roundtrip to the server to request assets on a webpage.
HTTP/2 Server Push is not advantageous in all situations, however, and can use excess bandwidth by pushing unneeded assets. Server Push can be deactivated using the “no push” directive header. Mobile networks (and other slow, lossy networks) are the most susceptible to both the benefits and downfalls of HTTP/2, as bandwidth and connection stability is constantly in flux.
By using the Canary version of Google Chrome, websites can test the performance of their site using HTTP/2 with and without Server Push. Safari, however, does not support Server Push and assets can only be pushed from the a single domain, not from a third party. “Uncacheable content”, “all content on a page” (unless these assets are already cached) and “the most likely next page” are examples of assets than can benefit most dramatically from the addition of HTTP/2 Server Push.