s3_algo/
config.rs

1use serde::{Deserialize, Serialize};
2#[derive(Clone, Debug, Serialize, Deserialize)]
3#[serde(default)]
4#[serde(deny_unknown_fields)]
5pub struct Config {
6    /// Maximum number of simultaneous upload requests
7    pub copy_parallelization: usize,
8
9    /// Maximum number of simultaneous download requests
10    pub download_parallelization: usize,
11
12    pub algorithm: AlgorithmConfig,
13
14    /// The "unit" of a delete request is number of objects
15    pub delete_requests: SpecificTimings,
16
17    /// NOTE: For now, `put_request` is used both in S3 `get`, `put` and `copy` operations.
18    /// Reason: We don't know if it's worth it with different configurations for these operations
19    /// that all have a duration that depends on the number of bytes of the objects in question.
20    /// The "unit" for such requests are number of bytes.
21    pub put_requests: SpecificTimings,
22}
23
24impl Default for Config {
25    fn default() -> Self {
26        Self {
27            copy_parallelization: 20,
28            download_parallelization: 20,
29            algorithm: Default::default(),
30            delete_requests: SpecificTimings {
31                seconds_per_unit: 0.2,
32                minimum_units_for_estimation: 10,
33            },
34            put_requests: SpecificTimings {
35                seconds_per_unit: 1.0 / 1_000_000.0, // 1 MBPS = 1e-06 seconds per MB
36                minimum_units_for_estimation: 10,
37            },
38        }
39    }
40}
41
42#[derive(Clone, Debug, Serialize, Deserialize)]
43#[serde(deny_unknown_fields)]
44pub struct AlgorithmConfig {
45    /// The base timeout which will always be there (an estimate of RTT)
46    pub base_timeout: f64,
47
48    /// Timeout is set to a fraction of expected upload time (> 1.0)
49    pub timeout_fraction: f64,
50
51    /// Every retry, the timeout is multiplied by backoff (> 1.0)
52    pub backoff: f64,
53
54    /// Number of times to retry a single request before giving up
55    pub n_retries: usize,
56
57    /// To estimate the upload speed incrementally, we use an exponential average:
58    /// `new_avg_speed = avg_power * new_speed + (1 - avg_power) * avg_speed`.
59    ///
60    /// Thus, between 0.0 and 1.0, closer to 1.0 means that newer data points have
61    /// more significance.
62    pub avg_power: f64,
63}
64impl Default for AlgorithmConfig {
65    fn default() -> Self {
66        Self {
67            base_timeout: 0.5,
68            timeout_fraction: 1.5,
69            backoff: 1.5,
70            n_retries: 8,
71            avg_power: 0.7,
72        }
73    }
74}
75
76/// These settings are specific to the kind of operation we do. For example delete or put in S3.
77#[derive(Clone, Debug, Serialize, Deserialize)]
78pub struct SpecificTimings {
79    /// The initial estimate of extra timeout per unit (byte or object)
80    pub seconds_per_unit: f64,
81    /// The amount of units in a request, below which it does not affect estimation
82    pub minimum_units_for_estimation: usize,
83}
84
85impl SpecificTimings {
86    /// Sane default setting for when the size is number of bytes
87    pub fn default_for_bytes() -> Self {
88        Self {
89            seconds_per_unit: 1.0 / 1_000_000.0,   // 1 MBPS
90            minimum_units_for_estimation: 500_000, // 500 KB
91        }
92    }
93    /// Sane default setting for when the size is number of objects
94    pub fn default_for_objects() -> Self {
95        Self {
96            seconds_per_unit: 0.2,
97            minimum_units_for_estimation: 2,
98        }
99    }
100}
101
102// DRAFT
103//
104// Now, we don't have "avg_min_bytes". Because... we will just substract the assumed constant
105// anyway.
106// What if the assumption is wrong?
107// Well, it should be rather small anyway. It is exclusively thought to be the RTT...
108// If the substraction is negative after all...? Then... idk
109
110// put_timeout_per_byte..?
111//  should configure it as an assumed MBPS just like before. expected_upload_speed.
112// delete_timeout_per_object..?
113//  quite straight-forward seconds per object
s3_algo/config.rs

s3_algo/
config.rs
