[cs615asa] meetup report

Wenyou Huang whuang12 at stevens.edu
Thu May 4 12:35:51 EDT 2017


Hello everyone,

I'd like to share some points from a  meetup on Wednesday, April 26, at
Facebook's office.

The first speech is about Linux Network Switches from Martín Beauchamp,
Shapeways on.  He mentioned GIFEE. GIFEE means Goole Infrastructure For
Everyone Else. 'EVERYTHING at Google runs in a container' Joe Beda, a
senior staff software engineer at Google. From the perspective, the GIFEE
has an intersection with the word.

In this speech, the speaker introduced some disadvantages in old network
topology(3 layers structure). As we know, Over the years, networks started
to use the "fat tree" model of connectivity using the core - distribution -
access architecture. However, The problem with traditional networks built
using the spanning-tree protocol or layer-3 routed core networks is that a
single "best path" is chosen from a set of alternative paths.  All data
traffic takes that "best path" until the point that it gets congested then
packets are dropped.

Thus, the speaker introduced a new topology, CLOS Topology. Clos networks
have now made their second reappearance in modern data center switching
topologies.

In Clos topology, every lower-tier switch is connected to each of the
top-tier switches in a full-mesh topology. If there isn't any
oversubscription taking place between the lower-tier switches and their
uplinks, then a non-blocking architecture can be achieved. The advantage of
the Clos network is you can use a set of identical and inexpensive devices
to create the tree and gain high performance and resilience that would
otherwise cost must more to construct. I think the point is suited for
distributed system. As we know, distributed system can lower the cost of
devices and gain not bad performance.


The second topic is about Automating The Linux Kernel Validation from
Yannick Brosseau, Facebook.

At Facebook, they push hard to always be running recent kernels. That have
many benefits but also means that we need to continuously validate these
new kernels.

In this presentation, he presented an overview of the kernel testing
infrastructure that they built that integrate many of the testing tooling
provided by the community. What's more, he presented the efforts of the
Linux community to improve the available testing tools and their
automation.

The last topic is about Dependency Traps for Microservices.

When you use the Internet, you contact with remote servers, and that will
return what you want. It seems like the remote server is a just machine,
you ask and it will give you answer. This is a black box thinking. But as
we know, the remote server is composed of two parts, Frontend, and Backend.
The front end is the interface of clients, which usually provide
interaction UI with clients. it's responsible for exchanging information.
Backend accepts clients' requests, and then deals with these different
requests,

There are three main section in backend, network access control list (ACL),
Encryption and Databases.
yes, in these sections, they also have their backend(client) and
frontend(server). In general, the frontend will deliver requests to the
backend, backend sends what it get to encryption section to decrypt the
contents from requests.

Got the requested content, the backend will check the ACL whether services
are allowed, if allowed, and then encrypt request and send it to databases.

The model looks like well, But Silvia presents some potential dependency
problems may come from the multiple servers. That demonstrate microservices
dependency design should not be overlooked.

She mentioned that dependency design must make sure:

1. Differentiate between bootstrap and runtime dependencies
2. Avoid local dependency cycles
3. Avoid bootstrap dependencied satisfied remotely.

At last, I almost understand the third presentation and the first
presentation, However, about the second presentation, I just master a part
of it. But whatever, I touch the frontier development, that's the point of
the meetup.

Thanks for your reading!

Wenyou Huang
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.stevens.edu/pipermail/cs615asa/attachments/20170504/e44eaedb/attachment.html>


More information about the cs615asa mailing list