Azure Storage Tiers: Optimizing Costs with Hot, Cool, and Archive

I wrote “Azure Storage Tiers: Optimizing Costs with Hot, Cool, and Archive” to share practical, production-minded guidance on this topic.

Understanding Access Tiers

Tier	Storage Cost	Access Cost	Retrieval Time	Use Case
Hot	Highest	Lowest	Immediate	Frequently accessed data
Cool	Medium	Medium	Immediate	Infrequently accessed (30+ days)
Archive	Lowest	Highest	Hours	Rarely accessed (180+ days)

Setting Blob Access Tier

# Upload blob to specific tier
az storage blob upload \
    --account-name mystorageaccount \
    --container-name mycontainer \
    --name myblob.txt \
    --file ./myfile.txt \
    --tier Cool

# Change existing blob tier
az storage blob set-tier \
    --account-name mystorageaccount \
    --container-name mycontainer \
    --name myblob.txt \
    --tier Archive

Using C#:

// C# - Managing blob tiers
using Azure.Storage.Blobs;
using Azure.Storage.Blobs.Models;

public class BlobTierManager
{
    private readonly BlobContainerClient _containerClient;

    public async Task UploadWithTierAsync(
        string blobName,
        Stream content,
        AccessTier tier)
    {
        var blobClient = _containerClient.GetBlobClient(blobName);

        var options = new BlobUploadOptions
        {
            AccessTier = tier,
            Metadata = new Dictionary<string, string>
            {
                { "originalTier", tier.ToString() },
                { "uploadedAt", DateTime.UtcNow.ToString("o") }
            }
        };

        await blobClient.UploadAsync(content, options);
    }

    public async Task ChangeTierAsync(string blobName, AccessTier newTier)
    {
        var blobClient = _containerClient.GetBlobClient(blobName);
        await blobClient.SetAccessTierAsync(newTier);
    }

    public async Task<AccessTier?> GetCurrentTierAsync(string blobName)
    {
        var blobClient = _containerClient.GetBlobClient(blobName);
        var properties = await blobClient.GetPropertiesAsync();
        return properties.Value.AccessTier;
    }

    public async Task BulkChangeTierAsync(
        IEnumerable<string> blobNames,
        AccessTier newTier)
    {
        var tasks = blobNames.Select(async name =>
        {
            var blobClient = _containerClient.GetBlobClient(name);
            await blobClient.SetAccessTierAsync(newTier);
        });

        await Task.WhenAll(tasks);
    }
}

Working with Archive Tier

# Python - Archive tier operations
from azure.storage.blob import BlobServiceClient, StandardBlobTier
from datetime import datetime, timedelta
import time

class ArchiveTierManager:
    def __init__(self, connection_string):
        self.blob_service = BlobServiceClient.from_connection_string(
            connection_string
        )

    def archive_blob(self, container_name, blob_name):
        """Move blob to archive tier"""
        blob_client = self.blob_service.get_blob_client(
            container_name, blob_name
        )

        # Set metadata before archiving
        blob_client.set_blob_metadata({
            'archivedAt': datetime.utcnow().isoformat(),
            'previousTier': blob_client.get_blob_properties().blob_tier
        })

        blob_client.set_standard_blob_tier(StandardBlobTier.ARCHIVE)
        print(f"Archived: {blob_name}")

    def rehydrate_blob(self, container_name, blob_name,
                       target_tier=StandardBlobTier.HOT,
                       priority='Standard'):
        """
        Rehydrate archived blob.
        Priority options:
        - 'Standard': Up to 15 hours
        - 'High': Under 1 hour (higher cost)
        """
        blob_client = self.blob_service.get_blob_client(
            container_name, blob_name
        )

        # Check current status
        properties = blob_client.get_blob_properties()

        if properties.blob_tier != 'Archive':
            print(f"Blob is not archived (current tier: {properties.blob_tier})")
            return

        # Start rehydration
        blob_client.set_standard_blob_tier(
            target_tier,
            rehydrate_priority=priority
        )

        print(f"Rehydration started for {blob_name}")
        print(f"Priority: {priority}, Target: {target_tier}")

    def check_rehydration_status(self, container_name, blob_name):
        """Check if blob is still rehydrating"""
        blob_client = self.blob_service.get_blob_client(
            container_name, blob_name
        )

        properties = blob_client.get_blob_properties()

        return {
            'blob_name': blob_name,
            'current_tier': properties.blob_tier,
            'archive_status': properties.archive_status,
            'is_rehydrating': properties.archive_status == 'rehydrate-pending-to-hot' or
                              properties.archive_status == 'rehydrate-pending-to-cool'
        }

    def wait_for_rehydration(self, container_name, blob_name,
                             timeout_hours=15):
        """Wait for rehydration to complete"""
        start_time = datetime.utcnow()
        timeout = timedelta(hours=timeout_hours)

        while datetime.utcnow() - start_time < timeout:
            status = self.check_rehydration_status(container_name, blob_name)

            if not status['is_rehydrating']:
                print(f"Rehydration complete. New tier: {status['current_tier']}")
                return True

            print(f"Still rehydrating... ({status['archive_status']})")
            time.sleep(300)  # Check every 5 minutes

        print("Rehydration timeout")
        return False

Cost-Based Tier Selection

// C# - Intelligent tier selection based on access patterns
public class TierOptimizer
{
    private readonly IMetricsService _metrics;

    public AccessTier RecommendTier(string blobName, BlobProperties properties)
    {
        var accessCount = _metrics.GetAccessCount(blobName, TimeSpan.FromDays(30));
        var lastAccess = _metrics.GetLastAccessTime(blobName);
        var blobSize = properties.ContentLength;

        // Calculate cost for each tier
        var costs = new Dictionary<AccessTier, decimal>
        {
            [AccessTier.Hot] = CalculateHotCost(blobSize, accessCount),
            [AccessTier.Cool] = CalculateCoolCost(blobSize, accessCount),
            [AccessTier.Archive] = CalculateArchiveCost(blobSize, accessCount)
        };

        // Consider minimum storage duration requirements
        var daysSinceLastTierChange = (DateTime.UtcNow - GetLastTierChange(blobName)).TotalDays;

        // Cool tier has 30-day minimum, Archive has 180-day minimum
        if (daysSinceLastTierChange < 30)
        {
            costs.Remove(AccessTier.Cool);
        }
        if (daysSinceLastTierChange < 180)
        {
            costs.Remove(AccessTier.Archive);
        }

        return costs.OrderBy(c => c.Value).First().Key;
    }

    private decimal CalculateHotCost(long sizeBytes, int monthlyAccesses)
    {
        decimal sizeGB = sizeBytes / (1024m * 1024m * 1024m);

        // Hot tier pricing (example - actual prices vary by region)
        decimal storageCost = sizeGB * 0.0184m;  // per GB/month
        decimal accessCost = monthlyAccesses * 0.0000004m;  // per operation
        decimal retrievalCost = 0;  // Free retrieval

        return storageCost + accessCost + retrievalCost;
    }

    private decimal CalculateCoolCost(long sizeBytes, int monthlyAccesses)
    {
        decimal sizeGB = sizeBytes / (1024m * 1024m * 1024m);

        decimal storageCost = sizeGB * 0.01m;
        decimal accessCost = monthlyAccesses * 0.00001m;
        decimal retrievalCost = sizeGB * monthlyAccesses * 0.01m;

        return storageCost + accessCost + retrievalCost;
    }

    private decimal CalculateArchiveCost(long sizeBytes, int monthlyAccesses)
    {
        decimal sizeGB = sizeBytes / (1024m * 1024m * 1024m);

        decimal storageCost = sizeGB * 0.00099m;
        decimal accessCost = monthlyAccesses * 0.00005m;
        decimal retrievalCost = sizeGB * monthlyAccesses * 0.02m;
        decimal rehydrationCost = monthlyAccesses > 0 ? 5.0m : 0;  // High priority rehydration

        return storageCost + accessCost + retrievalCost + rehydrationCost;
    }
}

Monitoring Tier Distribution

# Get blob count and size by tier
az storage blob list \
    --account-name mystorageaccount \
    --container-name mycontainer \
    --query "[].{name:name, tier:properties.blobTier, size:properties.contentLength}" \
    --output table

# Python - Generate tier distribution report
def generate_tier_report(container_client):
    tier_stats = {
        'Hot': {'count': 0, 'size': 0},
        'Cool': {'count': 0, 'size': 0},
        'Archive': {'count': 0, 'size': 0}
    }

    for blob in container_client.list_blobs():
        tier = blob.blob_tier or 'Hot'  # Default to Hot if not set
        if tier in tier_stats:
            tier_stats[tier]['count'] += 1
            tier_stats[tier]['size'] += blob.size

    # Format report
    print("\n=== Blob Tier Distribution Report ===\n")
    for tier, stats in tier_stats.items():
        size_gb = stats['size'] / (1024**3)
        print(f"{tier}:")
        print(f"  Blobs: {stats['count']:,}")
        print(f"  Size:  {size_gb:.2f} GB")
        print()

    return tier_stats

Best Practices

1. Analyze access patterns: Use metrics to inform tier decisions
2. Consider minimum storage durations: Avoid early deletion penalties
3. Plan for archive retrieval: Factor in rehydration time for workflows
4. Automate tier transitions: Use lifecycle management policies
5. Monitor tier-related costs: Track access and retrieval charges

Azure Storage tiers provide a powerful mechanism for balancing performance and cost, enabling organizations to store massive amounts of data economically while maintaining appropriate access times.\n\n## Takeaways\n\nAdd a concise, personal takeaway and recommended next steps here.\n