Skip to main content

Alexey Tarasevich

Alexey Tarasevich's Public Library

  • Human agency is the capacity for human beings to make choices

  • Non-firewalled clients and clients on publicly routable IP addresses can help NAT’ed nodes communicate by routing calls.
  • Calls are encrypted end-to-end
  • proxies with spare resources are chosen

3 more annotations...

  • A decentralized system is where some decisions by the agents are made without centralized control or processing.
  • In such a decentralized system, there is no single centralized authority that makes decisions on behalf of all the parties. Instead each party, also called a peer, makes local autonomous decisions towards its individual goals which may possibly conflict with those of other peers. Peers directly interact with each other and share information or provide service to other peers.
  • In decentralization the principle of subsidiarity is often invoked. It holds that the lowest or least centralized authority which is capable of addressing an issue effectively should do so.

  • it holds that social and political issues should be dealt with at the most immediate (or local) level that is consistent with their resolution

  • Замена абстрактных объектов и фактов
  • на понятия и представления, имеющие визуальное, аудиальное или кинестетическое представление
  • связывание объектов с уже имеющейся информацией в памяти

4 more annotations...

  • [pipeline:main] pipeline = catch_errors healthcheck cache authtoken keystone proxy-server [app:proxy-server] use = egg:swift#proxy account_autocreate = true [filter:keystone] paste.filter_factory = keystone.middleware.swift_auth:filter_factory operator_roles = admin, swiftoperator
  • $> sudo swift-ring-builder account.builder create 10 1 1 $> sudo swift-ring-builder container.builder create 10 1 1 $> sudo swift-ring-builder object.builder create 10 1 1
  • $> sudo swift-ring-builder account.builder add z1- 100 $> sudo swift-ring-builder container.builder add z1- 100 $> sudo swift-ring-builder object.builder add z1- 100

1 more annotation...

  • swift-ring-builder <builder_file> create <part_power> <replicas> <min_part_hours>
  • creating the <builder_file> with 2^<part_power> partitions
  • <min_part_hours> is the time in hours before a specific partition can be moved in succession (24 is a good value for this).

7 more annotations...

    • Define your policies in /etc/swift/swift.conf
    • Create the corresponding object rings
  • Once built or changed (rebalanced), you must distribute the rings to all the servers in the cluster.
  • swift-ring-builder <builder-file> remove <ip_address>/<device_name>

16 more annotations...

  • a bucket-level feature that enables automatic, asynchronous  copying of objects across buckets in different AWS regions

/srv 1

about 23 hours ago

  • services which require a single tree for readonly data, writable data  and scripts (such as cgi scripts) can be reasonably placed

  • the ring builder organizes the devices into tiers (first zone, then IP/port, then device ID)
  • When placing a replica, the ring builder looks for the emptiest device (biggest parts_wanted) in the furthest-away tier.
  • When generating handoff nodes, first the ring looks for nodes in other zones, then other ips/ports, then any other drive.

  • data is placed efficiently and with as much protection from hardware failure as possible
  • An availability zone is a distinct set of physical hardware with unique failure mode isolation.
  • distinct availability zones

7 more annotations...

  • a reduced redundancy storage policy
  • a geographically-specific storage policy
  • some data is trivially re-creatable and doesn’t require the same durability

5 more annotations...

  • major parts of global clusters are a region tier for placement, adjustable replica counts, a separate replication network, and read and write affinity
  • a region tier for data placement allows deployers to group availability zones into regions
  • Swift includes the ability to adjust the replica counts for existing cluster

2 more annotations...

  • creates a new market and value network and eventually disrupts an existing market and value network

  • companies can put too much emphasis on customers' current needs, and fail to adopt new technology or business models that will meet their customers' unstated or future needs

  • object storage has been built to scale as users consume an increasing amount of storage and want to access data from any device, anywhere.
  • every piece of data exists at the same level in a storage pool

  • disaster recovery is the main driver
  • applications in different datacenters all need low latency access to data
  • geographically dispersed sites, joined via high-latency network connections

17 more annotations...

May 26, 16

client --> proxy nodes --> storage nodes

container key problems: 
. a lot of shared keys to manage
. you need to do this on the per container basis
. many copies (e.g. 3 replicas per cluster)

container sync: short term fix, they are introducing regions tier to the replication
tiered zoning = "place data "as-uniquely-as-possible"

needed transition for customers: evolve small -> medium -> large deployments

primary + offsite regions: 
. "optimistically" create three replicas in the primary region (sync/async ?)
. move one to the offsite region asynchronously

when reading, options:
. get regionally local copy by default
. ask for the latest (proxy will query servers in other regions)

proxy have awareness of where it is tier-wise and can return closest replica

  • all the contents of a container can be mirrored to another container through background synchronization
  • operators configure their cluster to allow/accept sync requests to/from other clusters
  • the user specifies where to sync their container to

40 more annotations...

1 - 20 of 5801 Next › Last »
20 items/page

Highlighter, Sticky notes, Tagging, Groups and Network: integrated suite dramatically boosting research productivity. Learn more »

Join Diigo