This is an automated email from the ASF dual-hosted git repository.
juzhiyuan pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/apisix-website.git
The following commit(s) were added to refs/heads/master by this push:
new 606dd06 docs: update doc image (#936)
606dd06 is described below
commit 606dd069a7c5e9768fff8256dd0dac686583c569
Author: oil欧呦 <[email protected]>
AuthorDate: Fri Mar 4 18:12:49 2022 +0800
docs: update doc image (#936)
---
.../blog/2021/06/28/why-we-need-Apache-APISIX.md | 54 ++++++++++------------
1 file changed, 25 insertions(+), 29 deletions(-)
diff --git a/website/blog/2021/06/28/why-we-need-Apache-APISIX.md
b/website/blog/2021/06/28/why-we-need-Apache-APISIX.md
index 0b3d06e..cb59def 100644
--- a/website/blog/2021/06/28/why-we-need-Apache-APISIX.md
+++ b/website/blog/2021/06/28/why-we-need-Apache-APISIX.md
@@ -22,34 +22,32 @@ tags: [Technology]
Hello everyone, I'm happy to share a topic that I'm excited about, "Why do you
need Apache APISIX when you have NGINX and Kong".
-
-
The reason why we are doing a replacement project for NGINX and Kong is
actually related to the background of our backend architecture evolution, and I
will start by sharing with you the backend architecture evolution process,
which is very important.
I'll start by sharing with you the evolution of the backend architecture,
which is very important.
-
+
First of all, I would like to introduce myself, my name is Yuansheng Wang. I
wrote an e-book called "OpenResty Best Practices" in 2015 and formed a
community of over 10,000 people through this book. Since that time, I have
become more and more interested in open source itself. Before 2015, I was
basically mainly a user of open source software, then slowly became a
co-organizer of the community, and then later became a community leader.
Simple, because the book is written by you, others e [...]
-
+
I've been a teacher and eventually a community leader. In 2019, I founded
Shenzhen Tributary Technology Company with my partner Ming Wen, which is an
open source-based commercialization company. This company carries a lot of
personal ideals for both of us, and we can also say that we are doing the
ideals of every ordinary programmer, not wanting to be mediocre 996, I often
say to others that my dream is to "engrave my name into the history books", the
sad thing is that human beings no lo [...]
The sad thing is that mankind doesn't need a history anymore.
-
+
This is our team, we mainly collaborate remotely, and it's harder to get
everyone together. When there were only five or six people in the early stages
of the company, it was relatively easy to get the team together, but it hasn't
been together since this year, and this is the most together we've had so far
this year (but there are still a few students who didn't make it together).
As a technology-driven business company, technology has a very big say in our
company, and respect for technology starts with respect for technical talents.
There is no 996, no punching in and out, remote office, welcome interested
students to contact us, look forward to dreaming and ideal you to join our
company.
We are looking forward to your dream and ideal to join us.
-
+
The topic of this talk needs some background, so let's start with the history
of back-end architecture evolution. First, let's review this diagram, the right
part from top to bottom it is not a specific data flow diagram, it is the
history of our backend architecture evolution. Spring Cloud architecture mainly
serves JAVA language developers, Kubernetes is a container orchestration to
support any language, as well as the recent community hot topic service grid.
I often say to colleagues, let's look at the next five years, or even ten
years later, which architecture is the ultimate solution? From the current
information, the service grid will probably win. Even if it still has many
problems, I believe they can be solved.
-
+
At the beginning of the venture, it was particularly interesting to go through
this diagram in my head. We were able to see that as we gradually iterated on
the back-end architecture, we introduced a variety of different components. For
example, when we got to SOA, which is a service-oriented architecture, we
introduced reverse proxy components, usually NGINX and HAProxy, and when we
iterated to microservice architecture, we usually chose some more modern API
gateway products, such as Ko [...]
@@ -57,15 +55,15 @@ As the backend architecture continues to iterate and enters
the Kubernetes era,
Let's look at the left side of the more interesting JAVA, Spring Cloud
built-in API gateway has experienced ZUUL, ZUUL2, but still not good,
performance, architecture official are not satisfied, so Spring Cloud official
launched a new project Spring Cloud Gateway, the final formation of the family
bucket solution.
-Finally, the bottom right part of the service grid, for the service grid has
formed a choice istio (CP) + envoy (DP). Later we see the Alibaba open source
mosn, in a nutshell: Golang version of envoy.
+Finally, the bottom right part of the service grid, for the service grid has
formed a choice istio (CP) + Envoy (DP). Later we see the Alibaba open source
mosn, in a nutshell: Golang version of envoy.
-
+
Reviewing the previous architecture evolution diagram, I believe many students
have found out where the problem is. From top to bottom, from left to right,
for different scenarios, we finally "reasonably" introduced various components
to solve our problems, the architect's rule of survival: choose the most
suitable at the moment.
When we have few tools at hand, we always have to compromise between
functionality, dynamics, performance, etc. We have long been accustomed to and
even numb to the rapid development of IT technology, are they still the most
appropriate solution today?
-
+
As you can see, these are NGINX drawbacks, such as NGINX's low activity
community. While we could invest more resources at the corporate level, his
community is really unfriendly, and how unfriendly is it? As you can see in the
picture above, the NGINX repository in Github is only a mirror, the issue
function is closed, it is impossible to submit an issue, and even if you submit
a PR the official will not merge it.
@@ -73,13 +71,13 @@ In addition, NGINX is weak in its own routing, for example,
I want to do canary
Finally, the NGINX cluster management, almost every Internet vendor has its
own NGINX configuration management system, although the system is similar but
there is no unified solution, more than a decade has been blank.
-The system is similar but there is no unified solution, which has been blank
for more than ten years.
+The system is similar but there is no unified solution, which has been blank
for more than ten years.
Before talking further about Kong, I would like to talk to you about what is
cloud-native. This term has been around for a long time, but there is no
unified and clear definition until now. I synthesize several cloud vendors'
definitions and outline two main cloud-native features: first, it should
support containers, and second, it should support elastic and scalable
deployment. I think Kong does not fully meet the second, the official main
PostgreSQL relational database is a single poin [...]
The architecture is hard to choose.
-
+
Finally, a brief summary of the problems with NGINX and Kong.
@@ -101,25 +99,25 @@ In this architecture, you can't find a single point. Any
abnormal downtime of an
This is a diagram of the APISIX eco, from which you can see exactly what
peripheral ecologies are currently supported. On the left side are the
supported protocols, you can see the common Layer 7 protocols such as HTTP(S),
HTTP2, Dubbo, QUIC and IoT protocol MQTT, and the Layer 4 protocols such as
TCP/UDP. On the right are some open source or SaaS services such as SkyWalking,
Prometheus, Vault, etc. Below are some of the more common OS environments,
cloud vendors and hardware environment [...]
-
+
To give you a brief report on the current state of APISIX, APISIX has become
the most active open source API gateway project in the world in the two years
since it became open source, and this state has been going on for more than a
year. Remember the bottom sentence, APISIX has been **production available,
with better features, performance, and architecture across the board than
Kong**. In September 2019 Shell has already used the APISIX project in
production environments.
-
+
To briefly explain this graph, you can call it a contributor growth curve. The
horizontal coordinate is the timeline and the vertical coordinate is the total
number of contributors. We can see that APISIX and Kong are two relatively more
active projects. APISIX has been growing at a very good rate since the first
day of open source, and is growing at nearly twice the rate of Kong, which
shows how popular APISIX is. Of course there are many other ways to evaluate
the activity of a project [...]
-APISIX is still number one in these ways
+APISIX is still number one in these ways
After our actual customer visits, the feature of supporting multiple languages
is very necessary. After all, for many companies, they have their own familiar
technology stacks, and many companies are blank for NGINX C and Lua. APISIX has
officially announced multilingual support a few days ago, and currently
supports Java, and will gradually support Golang, Rust, NodeJS and other
languages.
APISIX's full dynamic and high performance is actually inseparable from the
high quality of the surrounding ecology. Other peripheral libraries such as
jsonschema, ipmatcher, etc. are several orders of magnitude better than similar
open source projects.
-
+
APISIX support for multi-language features have been put into the open source
project, interested students are welcome to follow and participate at any time.
The advantage of this implementation is that it is simple and universal, and
everyone can natively use their familiar language.
-
+
After all this talk, what are the advantages of APISIX for you? See the image
above.
@@ -129,31 +127,31 @@ High performance, dynamics, and an active community are
the trump cards of APISI
If one sentence sums up the pride of APISIX, I think it is:**APISIX, the most
active API gateway project in the world**. With this consensus, we tilt more
resources to the community, and we believe the community will make APISIX grow
soundly and healthily.
-
+
The APISIX goal: **Unified Proxy Infrastructure**.
You may be wondering if APISIX is going to support so many scenarios. Here I
will explain briefly that the core of APISIX is a high-performance proxy
service that does not bind any environment properties itself. When it evolves
into Ingress, Service Grid, etc., it is the external service that works with
APISIX, and it is the external program that changes rather than APISIX itself,
and we will explain how APISIX supports these scenarios step by step.
-
+
For traditional LB and API Gateway scenarios, APISIX has the advantage of
going from static to all dynamic, no more reloads, as many tech companies start
with a half hour NGINX reload. The aforementioned canary release scenario of
moduloing based on request id can be easily done in APISIX using fine-grained
routing.
-
+
-
+
-APISIX Ingress Controller solves all the problems mentioned above, and
inherits all the advantages of APISIX, in addition to supporting native k8s CRD
for easy migration.
+APISIX Ingress Controller solves all the problems mentioned above, and
inherits all the advantages of APISIX, in addition to supporting native K8s CRD
for easy migration.
-
+
Service mesh, it is necessary to talk to you about it. In the next five or ten
years, what is the most likely mainstream server-side architecture? If I were
to answer, I would choose the service mesh.
-
+
The diagram on the right shows the internal architecture of APISIX Mesh.
-
+
After talking so much about the present of APISIX, let's also talk about the
future of APISIX.
@@ -161,11 +159,11 @@ Because APISIX is currently an Apache Foundation project,
it is no longer the pr
The default configuration center for the open source version of APISIX is
etcd, and while it is still the best choice, we still often hear about support
for other configuration centers when we talk to users, more often than not
because etcd is so new that it is not on the list of supported products in the
company's existing operations and maintenance products. So we plan to make
APISIX work with other configuration centers.
-
+
APISIX is already on the road to full traffic data plane, and I believe we all
ask questions such as: Why do we need to unify traffic forwarding? Does
unification bring value to the enterprise? What are the benefits to the
technical staff? With these questions in mind, let's look at the following
diagram.
-
+
Unification itself is not the goal, but the benefits after unification is the
logic behind our pursuit, and several different perspectives are given below to
elaborate separately.
@@ -175,8 +173,6 @@ Unification itself is not the goal, but the benefits after
unification is the lo
- Company value: Unify technology stack, reduce company operation cost, reduce
the difficulty of transitioning to microservices and cloud-native, and
accelerate enterprise digital transformation.
-
-
Last but not least is the APISIX [Slack
channel](https://apisix.apache.org/docs/general/community#slack), any questions
can be left here or on [Github issue](https://github.com/apache/apisix/issues),
there will be someone to respond quickly, thanks again.
Click to watch the
[video](https://www.bilibili.com/video/BV1w54y1V73Z?p=1&share_medium=android&share_plat=android&share_source=COPY&share_tag=s_i×tamp=1621812452&unique_k=PEusrt)
