dataDistribution.py 6.43 KB
Newer Older
1
2
import random

Felix Seibert's avatar
Felix Seibert committed
3
import osd
4

5

Felix Seibert's avatar
Felix Seibert committed
6
7
8
9
class DataDistribution(object):
    """
    class to keep track of the osd (object storage device) locations of different folders, i.e.,
    their physical location.
10

Felix Seibert's avatar
Felix Seibert committed
11
12
    this class also allows to calculate several data distributions, e.g., mappings from folders to OSDs (each folder
    gets mapped to one OSD).
Felix Seibert's avatar
Felix Seibert committed
13
    """
14
15

    def __init__(self):
16
        self.OSDs = {}
17

18
    def add_new_osd(self, osd_uuid):
Felix Seibert's avatar
Felix Seibert committed
19
20
21
        """
        create a new empty osd and add it to the existing OSDs.
        """
22
23
        if osd_uuid in self.OSDs:
            print("key: " + osd_uuid + " is already present!")
24
            return
25
26
        new_osd = osd.OSD(osd_uuid)
        self.OSDs[osd_uuid] = new_osd
27

28
    def add_osd(self, new_osd):
Felix Seibert's avatar
Felix Seibert committed
29
30
31
        """
        add the given OSD (object) to the existing OSDs.
        """
32
33
        if new_osd.uuid in self.OSDs:
            print("key: " + new_osd.uuid + " is already present!")
34
            return
35
        self.OSDs[new_osd.uuid] = new_osd
36

37
    def add_osd_list(self, osd_list):
Felix Seibert's avatar
Felix Seibert committed
38
39
40
        """
        add the given list of OSDs (objects) to the existing OSDs.
        """
41
42
43
44
        for osd_uuid in osd_list:
            if osd_uuid not in self.OSDs:
                new_osd = osd.OSD(osd_uuid)
                self.OSDs[osd_uuid] = new_osd
45
46

    def get_osd_list(self):
Felix Seibert's avatar
Felix Seibert committed
47
48
49
        """
        get a list of all existing OSDs.
        """
50
        osd_list = []
51
        for osd_name in self.OSDs.keys():
52
53
54
            osd_list.append(osd_name)
        return osd_list

55
    def get_containing_osd(self, folder_id):
Felix Seibert's avatar
Felix Seibert committed
56
57
58
        """
        get the OSD containing the given folder_id, or None if the folder is not assigned to any OSD.
        """
59
60
61
        for checked_osd in self.OSDs.values():
            if checked_osd.contains_folder(folder_id):
                return checked_osd
62
63
        return None

64
    def get_average_folder_size(self):
Felix Seibert's avatar
Felix Seibert committed
65
66
67
        """
        get the average folder size of all folders of all OSDs.
        """
68
        total_size = 0
69
        total_number_of_folders = 0
70
71
72
        for one_osd in self.OSDs.values():
            total_size += one_osd.totalFolderSize
            total_number_of_folders += len(one_osd.folders)
73
74
        if total_number_of_folders == 0:
            return 0
75
76
77
        return total_size / total_number_of_folders

    def assign_new_osd(self, folder_id, new_osd):
Felix Seibert's avatar
Felix Seibert committed
78
79
80
        """
        assign folder_id to new_osd. if folder_id already is assigned to an OSD, this old assignment is deleted.
        """
81
82
83
84
85
86
        old_osd = self.get_containing_osd(folder_id)
        if old_osd is None:
            self.OSDs[new_osd].add_folder(folder_id, self.get_average_folder_size())
        else:
            self.OSDs[new_osd].add_folder(folder_id, self.OSDs[old_osd.uuid].folders[folder_id])
            self.OSDs[old_osd.uuid].remove_folder(folder_id)
87

88
    def add_folders(self, folders, osd_information=None, ratio_parameter='', random_osd_assignment=False):
Felix Seibert's avatar
Felix Seibert committed
89
90
        """
        adds a list of folders to the data distribution.
Felix Seibert's avatar
Felix Seibert committed
91
        if not specified otherwise, the assignments are calculated using the LPT algorithm.
Felix Seibert's avatar
Felix Seibert committed
92
93
94
95
        returns a list of assignments from folders to OSDs, for which (folders) there was previously no assignment.

        if the optional arguments are given, OSDs are assigned data proportionally to their ratio_parameter.
        """
96
97
        new_folders = []
        for folder in folders:
98
99
100
            containing_osd = self.get_containing_osd(folder.id)
            if containing_osd is not None:
                containing_osd.add_folder(folder.id, folder.size)
101
102
103
            else:
                new_folders.append(folder)

Felix Seibert's avatar
Felix Seibert committed
104
        print("dataDistribution: random_osd_assignment: " + str(random_osd_assignment))
Felix Seibert's avatar
Felix Seibert committed
105

106
107
        osds_for_new_folders = []

108
109
110
        # totally random OSD assignment, even ignoring OSD capacities
        # (might lead to I/O errors when too many groups are assigned to an OSD)
        if random_osd_assignment:
Felix Seibert's avatar
Felix Seibert committed
111
            print("using random osd assignment")
112
            for folder in new_folders:
Felix Seibert's avatar
Felix Seibert committed
113
114
                random_osd = random.choice(list(self.OSDs.values()))
                random_osd.add_folder(folder.id, folder.size)
115
                osds_for_new_folders.append((folder.id,
Felix Seibert's avatar
Felix Seibert committed
116
                                             random_osd.uuid))
Felix Seibert's avatar
Felix Seibert committed
117
            return osds_for_new_folders
118
119
120
121
122

        # balanced deterministic OSD assignment
        # (following largest processing time first or post-greedy approach)
        list.sort(new_folders, key=lambda x: x.size, reverse=True)

123
124
125
126
127
128
129
130
131
132
133
134
        osd_ratios = {}
        if osd_information is not None and ratio_parameter != '':
            total_osd_size = 0
            for osd_size in osd_information.values():
                total_osd_size += osd_size[ratio_parameter]
            for osd_uuid, osd_size in osd_information.items():
                osd_ratios[osd_uuid] = float(osd_size[ratio_parameter]) / float(total_osd_size)

        else:
            for osd_uuid in self.OSDs.keys():
                osd_ratios[osd_uuid] = float(1)

135
        for folder in new_folders:
136
137
138
139
140
141
142
            least_used_osd = None
            for one_osd in self.OSDs.values():
                if (least_used_osd is None) or \
                                        one_osd.totalFolderSize / osd_ratios[one_osd.uuid] \
                                <= least_used_osd.totalFolderSize / osd_ratios[least_used_osd.uuid]:
                    least_used_osd = one_osd
            least_used_osd.add_folder(folder.id, folder.size)
143
144
            osds_for_new_folders.append((folder.id,
                                         least_used_osd.uuid))
145

146
147
148
        return osds_for_new_folders

    def update_folder(self, folder, size):
Felix Seibert's avatar
Felix Seibert committed
149
150
151
        """
        updates the size of a given folder
        """
152
153
154
        for one_osd in self.OSDs.values():
            if folder in one_osd.folders.keys():
                one_osd.update_folder(folder, size)
155
156

    def description(self):
Felix Seibert's avatar
Felix Seibert committed
157
158
159
        """
        generates a string describing this data distribution
        """
160
        string = ""
161
162
        for one_osd in self.OSDs.values():
            string += str(one_osd)
163
            string += "\n"
164
            string += "folders : " + str(one_osd.folders)
165
            string += "\n"
166
        string += "average folder size: " + str(self.get_average_folder_size())
167
168
169
        return string

    def __str__(self):
170
171
172
173
174
        string_representation = "DataDistribution has " + str(len(self.OSDs)) \
                                + " osds: \n"
        for key, value in self.OSDs.items():
            string_representation += str(value) + " \n"
        return string_representation