forked from TieredMemDB/TieredMemDB
-
Notifications
You must be signed in to change notification settings - Fork 0
/
MANIFESTO
106 lines (95 loc) · 6.73 KB
/
MANIFESTO
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
[Note: this is the Redis manifesto, for general information about
installing and running Redis read the README file instead.]
Redis Manifesto
===============
1 - A DSL for Abstract Data Types. Redis is a DSL (Domain Specific Language)
that manipulates abstract data types and implemented as a TCP daemon.
Commands manipulate a key space where keys are binary-safe strings and
values are different kinds of abstract data types. Every data type
represents an abstract version of a fundamental data structure. For instance
Redis Lists are an abstract representation of linked lists. In Redis, the
essence of a data type isn't just the kind of operations that the data types
support, but also the space and time complexity of the data type and the
operations performed upon it.
2 - Memory storage is #1. The Redis data set, composed of defined key-value
pairs, is primarily stored in the computer's memory. The amount of memory in
all kinds of computers, including entry-level servers, is increasing
significantly each year. Memory is fast, and allows Redis to have very
predictable performance. Datasets composed of 10k or 40 millions keys will
perform similarly. Complex data types like Redis Sorted Sets are easy to
implement and manipulate in memory with good performance, making Redis very
simple. Redis will continue to explore alternative options (where data can
be optionally stored on disk, say) but the main goal of the project remains
the development of an in-memory database.
3 - Fundamental data structures for a fundamental API. The Redis API is a direct
consequence of fundamental data structures. APIs can often be arbitrary but
not an API that resembles the nature of fundamental data structures. If we
ever meet intelligent life forms from another part of the universe, they'll
likely know, understand and recognize the same basic data structures we have
in our computer science books. Redis will avoid intermediate layers in API,
so that the complexity is obvious and more complex operations can be
performed as the sum of the basic operations.
4 - We believe in code efficiency. Computers get faster and faster, yet we
believe that abusing computing capabilities is not wise: the amount of
operations you can do for a given amount of energy remains anyway a
significant parameter: it allows to do more with less computers and, at
the same time, having a smaller environmental impact. Similarly Redis is
able to "scale down" to smaller devices. It is perfectly usable in a
Raspberry Pi and other small ARM based computers. Faster code having
just the layers of abstractions that are really needed will also result,
often, in more predictable performances. We think likewise about memory
usage, one of the fundamental goals of the Redis project is to
incrementally build more and more memory efficient data structures, so that
problems that were not approachable in RAM in the past will be perfectly
fine to handle in the future.
5 - Code is like a poem; it's not just something we write to reach some
practical result. Sometimes people that are far from the Redis philosophy
suggest using other code written by other authors (frequently in other
languages) in order to implement something Redis currently lacks. But to us
this is like if Shakespeare decided to end Enrico IV using the Paradiso from
the Divina Commedia. Is using any external code a bad idea? Not at all. Like
in "One Thousand and One Nights" smaller self contained stories are embedded
in a bigger story, we'll be happy to use beautiful self contained libraries
when needed. At the same time, when writing the Redis story we're trying to
write smaller stories that will fit in to other code.
6 - We're against complexity. We believe designing systems is a fight against
complexity. We'll accept to fight the complexity when it's worthwhile but
we'll try hard to recognize when a small feature is not worth 1000s of lines
of code. Most of the time the best way to fight complexity is by not
creating it at all. Complexity is also a form of lock-in: code that is
very hard to understand cannot be modified by users in an independent way
regardless of the license. One of the main Redis goals is to remain
understandable, enough for a single programmer to have a clear idea of how
it works in detail just reading the source code for a couple of weeks.
7 - Threading is not a silver bullet. Instead of making Redis threaded we
believe on the idea of an efficient (mostly) single threaded Redis core.
Multiple of such cores, that may run in the same computer or may run
in multiple computers, are abstracted away as a single big system by
higher order protocols and features: Redis Cluster and the upcoming
Redis Proxy are our main goals. A shared nothing approach is not just
much simpler (see the previous point in this document), is also optimal
in NUMA systems. In the specific case of Redis it allows for each instance
to have a more limited amount of data, making the Redis persist-by-fork
approach more sounding. In the future we may explore parallelism only for
I/O, which is the low hanging fruit: minimal complexity could provide an
improved single process experience.
8 - Two levels of API. The Redis API has two levels: 1) a subset of the API fits
naturally into a distributed version of Redis and 2) a more complex API that
supports multi-key operations. Both are useful if used judiciously but
there's no way to make the more complex multi-keys API distributed in an
opaque way without violating our other principles. We don't want to provide
the illusion of something that will work magically when actually it can't in
all cases. Instead we'll provide commands to quickly migrate keys from one
instance to another to perform multi-key operations and expose the
trade-offs to the user.
9 - We optimize for joy. We believe writing code is a lot of hard work, and the
only way it can be worth is by enjoying it. When there is no longer joy in
writing code, the best thing to do is stop. To prevent this, we'll avoid
taking paths that will make Redis less of a joy to develop.
10 - All the above points are put together in what we call opportunistic
programming: trying to get the most for the user with minimal increases
in complexity (hanging fruits). Solve 95% of the problem with 5% of the
code when it is acceptable. Avoid a fixed schedule but follow the flow of
user requests, inspiration, Redis internal readiness for certain features
(sometimes many past changes reach a critical point making a previously
complex feature very easy to obtain).