Video summary

Lec-12: Storage in Cloud 🌧️| Microsoft Azure Storage 🗃️Services with Real life examples

Main summary

Key takeaways

Educational

Main ideas / concepts covered

  • Cloud storage purpose: Cloud storage is designed to handle very large volumes of data and provide:

    • Proper management
    • Performance
    • Scalability
    • Fast backup/recovery in case of disasters

  • Cloud storage strategy (data is not stored “directly” without planning):

    • When data arrives, you first determine what type of data it is.
    • Azure (as an example) provides different storage services tailored to different data types and usage patterns.

  • Azure storage service categories (as described):

    • (1) Object / image-video/binary-style storage: Used for data like images, videos, binary data (with different “varieties” in designed services).
    • (2) File storage: Used for text files and general files that need quick sharing/access.
      • Example: A multinational company where employees in different locations share files remotely via a repository.
    • (3) Queue storage: Used for messaging/requests that need asynchronous processing (to handle time delays).
      • Example: After a concert, a huge number of requests/messages arrive; queue storage stores them so processing can be done efficiently.
    • (4) Table storage: Used for schema-less / non-relational data, described as key-value pairs.
      • Example: IoT sensors (temperature/humidity) continuously generate data, stored using unique keys and corresponding values.
    • (5) Disk storage: Used for structured/sequential data using specialized structure (described as “structured disk storage” for sequential data).

  • Key principle: redundancy is essential

    • Data is not kept in only one place.
    • Azure architecture described as:
      • Region → multiple zones → multiple data centers
    • Policy mentioned: at least three copies within a region (and also copies across regions).
    • Rationale: ensures availability even if:
      • a data center fails,
      • a zone fails,
      • or an entire region faces a disaster.

  • Availability / “nines”

    • The lecture emphasizes extremely high availability (e.g., 999.999).
    • More nines = more important and reliable.

  • Tier-based storage (hot → cold → archive)

    • Not all data has the same importance/frequency.
    • Data is labeled by priority (examples: highest / VVIP / VIP / common).
    • Caching stores frequently requested data near where it’s needed for faster access.
    • Hot tier: accessed very frequently (e.g., most recent/high-demand data).
    • Cold tier: accessed less often (e.g., every 2–4 days or around 10 days).
    • Archive tier: rarely accessed (e.g., not used for a year).
    • Automatic movement described:
      • An email from a week/month ago moves from hot → cold.
      • If not accessed for ~6 months to a year, it moves to archive.

  • Storage media cost optimization

    • Higher-demand data uses faster/expensive storage (e.g., SSD).
    • Lower-demand data uses cheaper/denser media (e.g., HDD).
    • Archived data can use tape drives / magnetic tapes to reduce cost.

  • Generalization beyond Azure

    • The same conceptual approach applies to other cloud providers (AWS referenced): data-type-based storage selection, redundancy, and tiering remain core ideas.

Methodology / instruction-like steps (detailed)

  1. On incoming data:

    • Identify data type/variety (e.g., images/videos/binary, files/text, messages/requests, key-value style data, structured sequential data).

  2. Choose the appropriate storage service based on data type:

    • Use the corresponding category:
      • object-style for media/binary,
      • file storage for shared files,
      • queue storage for async messages,
      • table storage for schema-less key-value data,
      • disk storage for structured/sequential data.

  3. Ensure redundancy by design:

    • Store multiple copies:
      • At least three copies within the same region,
      • plus additional copies across zones and other geographic regions (intra- and inter-region redundancy).

  4. Apply tiering based on access frequency:

    • Label data by priority/importance.
    • Keep frequently accessed data in hot tier.
    • Move moderately accessed data to cold tier.
    • Move rarely accessed data to archive tier.

  5. Use caching to speed access for high-frequency data:

    • Cache data on the nearest server to reduce response time.

  6. Match storage media to demand to control cost:

    • Use SSD for hot/high-demand data,
    • hard disks for medium-demand data,
    • tape drives/magnetic tapes for archive/low-demand data.

  7. Aim for high availability (“nines”):

    • Maintain infrastructure and redundancy so the service remains available even during disasters.

Speakers / sources featured

  • Speaker: The lecture narrator / instructor (e.g., “Dear students…”, “I am going to explain…”, “from Azure’s point of view…”)
  • Source/technology mentioned: Microsoft Azure (primary example) and AWS (mentioned as an analogy/generalization)
  • No other specific named individuals or organizations were featured.

Original video